An Internship Report on

A Panel Data Study on Assessing the Influence of Digitalized Financial Services on the
Performance of Banks: A Case Study on the Selected Commercial Banks of Bangladesh

Submitted To
Department of Banking and Insurance
Faculty of Business Studies
University of Dhaka

Supervised By
Abdullah Al Mahmud, PhD
Professor
Department of Banking and Insurance
Faculty of Business Studies
University of Dhaka

Submitted by
Sorna Akter
Section- A
ID- (24-080)
Department of Banking and Insurance
University of Dhaka

Date of submission: 5th January, 2023.

 Letter of Transmittal
5th January, 2023

Professor,

Abdullah Al Mahmud, PhD

Department of Banking and Insurance,

Faculty of Business Studies,

University of Dhaka.

Subject: Submission of Internship Report on “A Panel Data Study on Assessing the Influence of
Digitalized Financial Services on the Performance of Banks: A Case Study on the Selected Commercial
Banks of Bangladesh”
Sir,
It is my great pleasure that finally I can submit my internship report on “A Panel Data Study on Assessing
the Influence of Digitalized Financial Services on the Performance of Banks: A Case Study on the
Selected Commercial Banks of Bangladesh” as per the academic requirement for completing the BBA
Program from the Department of Banking and Insurance.
I am highly grateful to you that you have assigned me such great work which has helped me to gain and
enrich my practical knowledge regarding Fintech banking Services.
I have given my best effort to complete this study. I have gained vast practical knowledge from this
internship Program. Therefore, I hope that this study will be able to meet its standard and service purpose
accordingly.

Regards,

…………………………………

Sorna Akter
Id- 24-080
Section: A
Batch: 24th

Page |
 Student’s Declaration

I, Sorna Akter with ID No: 24-080, hereby declare that I am submitting my internship report on “A Panel
Data Study on Assessing the Influence of Digitalized Financial Services on the Performance of Banks: A
Case Study on the Selected Commercial Banks of Bangladesh”. In this report, the situation of financial
technologies and the factor that affect it are identified which is solely developed by myself. I was
thoroughly supervised by my internship supervisor honorable Professor Abdullah Al Mahmud, (Ph.D.),
Department of Banking and Insurance, University of Dhaka while preparing this report. I am also
declaring that I have not shared this report with any other platform or individuals and I have not copied
any content from anyone else’s work. I am also declaring myself liable for any kind of breach of the
academic regulations for preparing an internship report.

.............................................................

Sorna Akter

ID: 24-080

Section A

BBA 24th Batch

Department of Banking and Insurance

University of Dhaka

Page |
 Certificate of Supervisor

This is to certify that Sorna Akter, a student of the Bachelor of Business Administration (BBA), 24 th
batch, ID: 24-080, Department of Banking and Insurance at the University of Dhaka is under my
supervision as the partial fulfillment for the award of BBA degree. She has completed the work under my
direction and supervision. The completion of this report has been a major focus for her. I pray that she
achieves success and prosperity.

……………………………………………

Abdullah Al Mahmud, PhD

Professor

Department of Banking and Insurance

University of Dhaka

Page |
 Acknowledgment

I would like to express my gratitude and indebtedness to my honorable teacher and supervisor (Professor,
Abdullah Al Mahmud, Ph.D. Department of Banking and Insurance, University of Dhaka). I was able to
complete this job effectively due to his unlimited direction, invaluable counsel, ongoing inspiration,
constructive criticism, and kindness. I would also want to thank the journalists and web developers whose
publications and websites assisted me in gathering all the required material. I would also want to thank
everyone especially my friends, classmates, and roommate who provided insightful feedback and helped
me produce a successful thesis. Finally, I would like to express my gratitude to all other fellow mates who
directly or indirectly assisted me in providing and compiling the essential data for the completion of this
report.

Page |
 Executive Summary

Among the many subsectors of the financial services industry, banking was an early adopter of rapid
globalization and a major beneficiary of advances in information technology. Bangladesh's private banks
have been making efforts as of late to adopt the financial system of more industrialized nations. The
Dhaka Stock Exchange enlisted 10 private commercial banks have been taken to show the impact. The
information for this study is taken from the respective banks' annual reports and online presence. The
positive and negative effects of technology on the profits of the banking industry can be learned through
an analysis of its effect on the private commercial banks in Bangladesh. In recent years, it has been
common practice for commercial banks in Bangladesh to use some kind of automation, even only at the
central office. Statistics on automated teller machines, virtual banking branches, mobile banking, and
internet banking penetration are also included. Bank branch automation is lagging behind the fast
expansion of ATM availability. One of the newest trends in our nation is mobile banking, which has been
shown to increase profitability. Technology use was not crucial to survival in the previous time period.
But recent research has shown that technologically proficient institutions earn more profits than their less
innovative counterparts. For the panel data analysis, the empirical model consists of pooled OLS model.
The best-performing model in the analysis was identified using this test. Where this is evident, consumers
have shown a heightened interest in the availability of ATMs and banking options over the Mobile and
Internet.

Page |
 Table of Contents
Chapter 1: Introduction..................................................................................................................................1

1.1 Introductory Overview.........................................................................................................................1

1.2 Background of the Study.....................................................................................................................1

1.3 Rationale of the Study..........................................................................................................................2

1.4 Research Questions..............................................................................................................................3

1.5 Objectives............................................................................................................................................3

1.6 Conceptualization................................................................................................................................3

1.6.1 Digitalization of Banks.................................................................................................................3

1.6.2 Fintech Banking............................................................................................................................3

1.6.3 Mobile Financial Service (MFS)..................................................................................................4

1.6.4 Payment System Operator (PSO)..................................................................................................4

1.6.5 Point of Sale..................................................................................................................................4

Chapter 2: Literature Review.........................................................................................................................5

Chapter: 3 Empirical Methodology.............................................................................................................10

3.1 Sample and Data................................................................................................................................10

3.2 Variables............................................................................................................................................10

3.3 Research Model.................................................................................................................................11

3.4 Limitations of the Study.....................................................................................................................12

Chapter 4: Data Analysis.............................................................................................................................13

4.1 Regression Model Analysis...............................................................................................................13

4.1.1 ROA as a Dependent Variable....................................................................................................13

4.1.2 ROE as a Dependent variable.....................................................................................................18

4.1.3 NIM as A Dependent variable....................................................................................................23

4.1.4 BEP as A Dependent Variable....................................................................................................27

Page |
 4.2 Trend Analysis...................................................................................................................................31

4.2.1 Numbers of Agents.....................................................................................................................31

4.2.2 Numbers of Registered Customers.............................................................................................31

4.2.3 Numbers of Effective Account...................................................................................................32

4.2.4 Amount of All Transactions........................................................................................................32

4.2.5 Total Number of ATMs..............................................................................................................33

4.3 Comparative Analysis........................................................................................................................33

4.3.1 Areas of service comparison.......................................................................................................33

4.3.2 Total Subscribers........................................................................................................................34

4.3.3 ATMs..........................................................................................................................................34

4.3.4 Agents.........................................................................................................................................35

4.3.5 Branches......................................................................................................................................35

Chapter 5: Findings......................................................................................................................................37

5.1 Findings of Regression Result...........................................................................................................37

5.2 Findings of Trend Analysis................................................................................................................38

5.3 Findings of Comparative Analysis.....................................................................................................38

Chapter 6: Conclusion..................................................................................................................................39

6.1 Conclusion.........................................................................................................................................39

6.2 Prospects of Fintech Banking in Bangladesh.....................................................................................39

6.3 Recommendations..............................................................................................................................40

References....................................................................................................................................................42

Page |
 Lists of Tables
Table 1: A list of the variables and associated definitions...........................................................................12
Table 2: Output of model coefficients, incorporating (Pooled Ordinary least square (OLS), Fixed effect
(fe), Random effect (re), Generalized Least square (GLS))........................................................................14
Table 3: Output of Hausman Test for ROA.................................................................................................15
Table 4: Output of B/P LM Test for ROA...................................................................................................15
Table 5: Multicollinearity Test for ROA.....................................................................................................16
Table 6: Heteroscedasticity Test for ROA...................................................................................................17
Table 7: Wooldridge Test of Autocorrelation for ROA...............................................................................17
Table 8: RAMSEY Reset Test for ROA......................................................................................................18
Table 9: Output of model coefficients, incorporating (Pooled Ordinary least square (OLS), Fixed effect
(fe), Random effect (re), and Generalized Least square (GLS)) for ROE...................................................19
Table 10: Output of Hausman Test for ROE...............................................................................................20
Table 11: Output of B/P LM Test for ROE.................................................................................................20
Table 12 Multicollinearity Test ROE..........................................................................................................21
Table 13: Heteroscedasticity Test for ROE.................................................................................................21
Table 14: Wooldridge Test for Autocorrelation for ROE............................................................................22
Table 15: RAMSEY Reset Test for ROE....................................................................................................22
Table 16: Output of model coefficients, incorporating (Pooled Ordinary least square (OLS), Fixed effect
(fe), Random effect (re), and Generalized Least square (GLS)) for ROE...................................................23
Table 17: Output of Hausman Test for NIM...............................................................................................24
Table 18: Output of B/P LM Test for NIM..................................................................................................25
Table 19:Multicollinearity Test for NIM.....................................................................................................25
Table 20: Heteroscedasticity Test for NIM.................................................................................................26
Table 21: Wooldridge Test of Autocorrelation for NIM.............................................................................26
Table 22: RAMSEY Reset Test for NIM....................................................................................................26
Table 23: Output of model coefficients, incorporating (Pooled Ordinary least square (OLS), Fixed effect
(fe), Random effect (re), and Generalized Least square (GLS)) for BEP....................................................27
Table 24: Output of Hausman Test for BEP................................................................................................28
Table 25: Output of B/P LM Test for NIM..................................................................................................29

Page |
Table 26: Multicollinearity Test for BEP....................................................................................................29
Table 27: Heteroscedasticity Test for BEP..................................................................................................30
Table 28: Wooldridge Test of Autocorrelation for BEP..............................................................................30
Table 29: RAMSEY Reset Test for BEP.....................................................................................................30

Page |
Chapter 1: Introduction
1.1 Introductory Overview

Today’s modern world is the result of some major revolutionary breakthroughs in the area of technology.
It continues moving to the road of digitalization in every aspect of life. These life-evolving innovations
have put footsteps also in the development of the financial sector. Digitalization is a huge concept that, in
the banking industry, means using Information and Communication Technology (ICT) to change the
traditional banking model into a more sophisticated, value-producing, and advanced business framework.
Financial technology (fintech) is the adoption of digital breakthroughs by financial institutions in
developing financial products and financial service delivery Fintech is continuously spreading its benefits
to both financial service providers and customers. The financial institutions of Bangladesh have also
started the application of financial technology and within a short time, the financial landscape of this
sector has been significantly changed. Mobile Financial Services (MFS), Payment Service Operator
(PSO), Payment Service Provider (PSP), Automated Clearing House (ACH), Automated Cheque
Processing (ACP), Automated Fund Transfer Networks (AFTN), Real Time Gross Settlement (RTGS),
etc. have made the transactions, payment system, fund collection, fund disbursement, and overall banking
operations more swift, precise and efficient.

1.2 Background of the Study

By combining the phrases digital technology and financial services the name "FinTech" originated. To
put it briefly, FinTech generally fosters companies to amputate digital technology to develop products and
services including mobile payments, online banking, alternative finance, general financial management,
and big data (Zveryakov et al., 2019). People's mistrust of the former banking system grew throughout
2008-2014 as they learned the causes of the International Finance Disaster, which quickly ballooned into
a wider economic calamity (Lee, 2017). Some banks operators losing their jobs and the resulting shift in
public opinion paved the way for the development of Fintech 3.0. This age is also marked by the
emergence of new actors, most notably fintech banks, who have joined the traditional incumbents.
Introduction of Bitcoin v0.1 in 2009, which was followed by the growth of other cryptocurrencies and
then a huge crypto meltdown, was another event of significance in the financial world (Wahab, 2021).
Fintech 3.5 ushers in a departure from the western financial system by taking into account the worldwide
proliferation of digital banking made possible by technological advances in fintech. It concentrated on
consumer behavior and how individuals from developing countries adopt the internet. For example,

Page |
markets in India and China that had not the opportunity to make the same types of physical financial
services as Innovative ideas were more likely to be accepted by Western markets (Zhao,2019). Before
2010, Bangladesh's fintech landscape mostly consisted of credit and debit cards, ATMs, point-of-sale
terminals, and to a smaller degree, the banks' internet banking services. In Bangladesh, Dutch Bangla
Bank became the first bank to launch online banking in 2003 (Taher, et al., 2022). Despite that access to
financial services was not widely available. Bangladesh's financial inclusion rate was just 32% in 2011, as
reported by the Global Findex Database of the World Bank. When Bangladesh Bank initially launched
the Bangladesh Electronic Funds Transfer Network (BEFTN) in 2011 to increase the flexibility of
electronic payment systems and mobile financial services, the fintech landscape in Bangladesh started to
alter. To improve bank interoperability, National Payment Switch Bangladesh (NPSB) was introduced in
2012. In 2013, agent banking was launched as a continuation of MFS. Bangladesh Bank and A2i worked
together to establish the Digital Financial Services (DFS) Lab in the nation for the development of the
fintech ecosystem. Using mobile technology, the use of fintech is steadily growing in Bangladesh.

1.3 Rationale of the Study

This study will assist to identify the possible impacts that originated from using and adopting new
technology and digital transformation of banking operations by the banks of Bangladesh. The utilization
of mobile banking services, an online efficient payment system, and proficient banking operations have
many significant aspects. Banks with new technology can magnify financial inclusion drastically that is
more people, especially the rural poor folk being gathered under the umbrella of banking, and the
profitability, stability, and competitiveness of the banks are also considerably influenced. Customers
expect banking to be easy, seamless, and pleasant in the modern digital era. They also expect their banks
to offer them cutting-edge, innovative goods and services. Therefore, banks with digital transformation
can attract more customers and be efficient in making customer satisfaction and customer retention This
study will be pursued by taking the profitability measures, operational efficiency indicators, and
competitive aspects. therefore, it will help to measure the possible effects of these financial technologies
and digital operations on banks’ profitability as well as on banks’ overall performances. Moreover, the
theoretical overview of fintech banks, their current status in the banking industry of Bangladesh, their
comparative situation with traditional banks, and their prospects and possibilities in these technological
fields can be thoroughly assumed from this study.

Page |
1.4 Research Questions

Some major findings are to be conducted throughout the whole research paper to get the conclusion on
such queries. The research questions are discussed as follows.

 What impacts does fintech have on banks’ profitability and financial stability?
 How does digitalization increase banks’ operational efficiency?
 In what way do digitalization and fintech affect banking Competitiveness?

1.5 Objectives

Many banking companies start using mobile financial services and almost all of the banks in Bangladesh
adopt online banking and payment services. The general objective of this thesis is to explore all the
aspects of digital transformation and financial technologies in the field of the banking sector operating in
Bangladesh. The specific objectives to gain the general aim of this report are given as follows.

 To find out the possible impacts of digitalization and financial technologies on the performance
of banks.
 To assess the influences of these digital systems and online platforms of banking on the operating
capacity, management skills, and overall operating performance.
 To know the contribution of the adoption of these technologies on the banks’ return on asset,
return on equity (indicators of profitability), and discernment of how net interest margin.

1.6 Conceptualization
1.6.1 Digitalization of Banks
Online banking is just one aspect of the virtual process known as "digital banking." Digital banking
includes the upfront end that customers identify, the backward end that bankers watch via the servers,
conductor control panels, and the middleware that links the nodes as a closing platform (Nayak, 2018). A
digital bank must, in the end, enable all banking functionality throughout all stages for service delivery. In
another way, it has the ability to perform all the functions of a principal office, branches, an online
service, a card system, ATM facilities, and point-of-sale (POS) tools.
1.6.2 Fintech Banking
Fintech, as it is more often known, dictates innovative technology that intends to enhance and automate
the offering of financial services. Fintech is used by banks for both consumer-facing solutions such as the
app being used to monitor one's account as well as back-end operations like tracking account activity in

Page |
the background (Thakor, 2020). Fintech is used by banks as well to underwrite loans. Fintech allows
people to use numerous bank services, such as paying with a smartphone for purchases and getting
financial advice on home computers. Banks employ fintech for both front-facing solutions, like the app
used to track one's account, and back-end tasks, such as monitoring financial accounts in the background.
Banks also utilize fintech to evaluate loan applications. People can use a variety of bank services thanks
to fintech, including making purchases using their smartphones as payment and receiving financial advice
on their home computers.
1.6.3 Mobile Financial Service (MFS)
Mobile financial services (MFS) is a process of providing banking services using wireless mobile
networks that facilitates the customer to conduct transactions and get other financial assistance through a
mobile phone. It provides several facilities including the deposit of funds into the respective accounts,
withdrawal of money, and transferring and receiving of funds (Kim et al., 2018). These services are
generally provided with the assistance of bank agents, enabling mobile bank customers to operate
transactions at unaffiliated agent sites other than bank offices. With the idea of a "mobile wallet" account,
MFS incorporates services like payment processing and mobile banking that offer convenience and
security for transfers, payments, and savings (Mallat et al.,2018).

1.6.4 Payment System Operator (PSO)

An organization in charge of running a payment system is known as a payment system operator. The PSO
uses certain operating models to deliver services. They generally delegate the payments and settlement-
concentrated tasks to different other organizations (Chapman et al., 2017). A person or organization that
has been granted a license by the Bangladesh Bank to operate a settlement for payment activities with
both or among participants, where the primary participant has to be a scheduled bank and financial
institution that maintains accounts with the Bangladesh Bank in order to comply with the Cash Reserve
Requirements, is referred to as a "Payment System Operator".

1.6.5 Point of Sale

A point of sale, sometimes known as a POS, is a machine that handles retail customers' transactions. For
businesses to accept payments made with credit or debit cards, POS terminals must be installed
(debit cards, credit cards, smart cards, etc.) (Vukadin, et al., 2016). The installation of a POS device at a
retail location will minimize cash handling and increase sales for the retailer by allowing for more
methods of payment. Customers can purchase without worry because they don't need to bring cash with
them.

Page |
Chapter 2: Literature Review

A study, by (Hossain et al., 2021) shows how the use of electronic banking applications affects the overall
performance of organizations, and how financial technologies assist in core banking solutions. The work
has been performed on four state-owned commercial banks in Bangladesh. The model has been developed
by taking a panel dataset of the sample banks using the ordinary least square method (OLS). Profitability
measures, the concept of e-banking, statistical calculations, and diagnostic test are incorporated into the
model to measure the contribution of technology to the banks’ performance. The main research gap and
limitation of this study is the taking of a very few sample size which is only four banks of Bangladesh and
thus lesser data analysis. This can be exalted by taking both the public and private sector banks
functioning in Bangladesh.

(Wang et al., 2021) conducted research that focused on the influence of the undertaking of financial
technology on banks’ profitability. This paper also exhibits in what way the technologies help reduce
banks’ credit and operational risk. Incorporating 113 sample sizes with a time duration of 2008-2017, the
study was on domestic commercial banks in China. The endogenous complexity is eliminated with the
use of Differential Generalized Moment Estimation (DIF-GMM) and Systematic Generalized Moment
Estimation (SYS-GMM) techniques. The methods that were also included are the multicollinearity
matrix, baseline regression, and descriptive data. The limitation of this paper is the incorporation of only
the domestic banks which can be elevated by analyzing the private and non-domestic banks also.

(Wu et al. 2021) mentioned in his research the adoption and implementation of ICT by state-owned banks
in China and attempts to find out the impact of digitalization and technology-based financial product
design on profitability as well as on the leverage of banking services. The analysis is based on state-
owned commercial banks with a sample size of six from 2014 to 2019. OLS model, FGLS estimation, FE
estimation, and Paris-Weinstein estimation have been adopted to ensure statistical inference. Diagnostic
tests, implementation of panel data, and comparative analysis of estimators have also been accomplished.
Lesser data duration is the main research gap of this paper as the introduction of advanced technologies
can significantly change the impacts in recent years.

(Chindudzi et al. 2020) studied the commercial banks of Zimbabwe to find out how digitization affects
the efficiency of the banks. The main goal of this paper is to figure out how advanced technology can
speed up the financing process and lower operational risk. Analyzing the financial annual reports and
annual publications, the four variables have been collected on digital banking having a time duration of
2013-2017. The explanatory variables were the number of electronic transactions, online fees, and

Page |
charges, deposits collected from the online platform, and expenditure incurred for online banking, and the
explained variable is the ROA as the indicator of the financial performance. Pearson's product-moment
creation analysis and panel data analysis were used as statistical analytical tools.

In a research (Arrawatia et al., 2014), shows the relationship between the adoption of financial
technology, banks’ performance, and banks’ competitiveness. Taking the period 1996 to 2011, the
granger causality test was applied to examine the assets of the banks to determine the banks' proficiency
and competitiveness. The results of the paper showed a rising trend in competitiveness and demonstrated
digitalization influence competitiveness and, competitiveness promotes efficiency. Most of the research
papers on financial technology cover the basic terminologies like why fintech should be applied in a firm,
how it can be implemented, and its basic applications.

In his research (Chen, 2021) explores how the performance of Chinese banks is impacted by the
implementation of fintech programs. The scenario of customer retention and customer satisfaction on
basis of the application of technologies is also explained in the study. Financial service excellency and
operating efficiency are the indicators of the banks’ performance thus while adopting fintech, these
aspects should be of concern. The primary information is collected in the form of questionnaires and then
prepared in a model equation to study. The study results have positive outcomes regarding the
contribution of financial technology to customers' pleasantness. PD under service technologies had a
negatory effect on customer amusement and expectation of assistance. Standard service quality and high
efficacy of management can mitigate the negative aspects highlighted regarding FTPs.

In a study, (Sufian and Habibullah, 2010) examine the influences of bank-specific and macroeconomic
factors on the profitability of the banks of Thai banks taking the period from 2010 to 2020. After running
a regression model, the study found that there is a significant positive relationship between bank size,
profitability, and capitalization. However, operating costs, non-interest revenue, and the non-performing
loan had a negative relation with the bank's profitability. Though NPL had positive influences on ROE, it
has opposite relation with ROA. The impact of large economic advancement had a positive response to
the inflation of the banks of Thailand whereas the impact of GDP on the performance of the banks was
negative.

(Onay and Ozsoz, 2013) conducted a research to assess the influences of e-banking on the overall
performance of commercial banks in Turkey. The total sample size was 18 commercial banks operating in
Turkey. Bank-specific factors and the qualitative information on the adoption of fintech banking were the
explanatory variables and the overall performance of the bank was denoted as an independent variable.

Page |
Conducting regression using panel data, the study results in a positive effect of the implementation of e-
banking on the operating efficiency of the taken banks. However, the interest revenue of the banks was
seen to decline due to increased competition in the industry.

In a paper (Hernando and Nieto, 2007) examine the effect of digital payment mechanisms on the
economic functions of the banks of Span. A total of 72 commercial banks of Span were taken as a sample
to conduct the study for the period of 1999 to 2006. There was a positive finding that the adoption of new
technology for transaction delivery functions has a favorable outcome on the profitability of the sample
banks. The study concludes that these online payments and digital transactional tools can be used as
supplementary mechanisms other than the implantation of ATMs or bank offices.

In a study, (Aduda and Kingoo, 2012) associated e-banking operations and financial activities building a
positive relationship between these two sectors of the commercial banks of Kenya. The Pearson Product-
Moment Correlation Coefficient model was used to detect the connection between these factors. The
model test determined favorable contracts for the implementation of digital banking activities with the
operational performance of the sample banks and the effects are significant.

(Haller and Siedschlag, 2011) discovered that comparatively smaller companies are more effective in
using and implementing financial technologies than large firms. To assess the contribution of these
technologies to the banks’ profitability, a large range of variables which are known as control variables
related to the bank are used. The result shows the presence of a strong connection between the utilization
of fintech and the evolvement of banks’ efficiency. A sensitivity test accompanying the effects of the
global financial crisis and the determinants of several panel data was conducted. The results of these other
tests show the negative impact of these technologies on the performance of the banks.

In a research paper, (Gozman et al., 2018) highlights the importance of fintech startups in addressing
customer complaints. After analyzing prospects and probabilities, it concluded that new financial models
are able to give solutions to the problems of the customer in a more innovative, efficient, and cost-
effective way. It also exhibits that by combining financial activities and financial technologies, even small
startups can compete with large traditional banking institutions in the aspects of customer inclusion,
customer satisfaction, customer retention, and cost reduction.

(Tao et al., 2022) explains the role of financial technologies in explaining the environment from the
perspective of banking and the financial platform is shown in the paper. Using the lending cap indicating
peer-to-peer distribution, fintech banks can cover those regions that are not reached by traditional banking
activities. By playing an influential role and reaching the unserved areas by the traditional banks, fintech

Page |
banks create effective competition with conventional banks. With this growing competition and larger
market capture, these startups snatch a portion of the profit of traditional banks.

A study, (Katsiampa et al.,2022) examines the influences of new financial technologies on the overall
financial performance of commercial banks in China. It found that the adoption of financial technologies
adversely reduces the base profit of Chinese banks and thus declines the strongness of financial
performances. The banking sector continuously faces financial distresses due to ever-changing economic
cycles and these changing cycles are influenced by the adoption of new technologies. To remove these
drawbacks, the banking sector requires a place to grow up, and some innovations like blockchain to make
payment transparency. Furthermore, they need to strategically adopt financial technology to get
advantages from these innovations.

In a research paper, by (Gomber et al., 2018) brings the aspects of how a conventional company can be
moved getting the merger and acquisition options with larger banks using financial technologies. Many
banks are merged or get acquisition trying to get benefits from the principal company that adopts
digitalization in operation and financial technologies in services. But there an argument arises regarding
whether these mergers and acquisitions are made to reduce the competition in the market. In a general
sense, these takeovers are made to make growth, be successful, and sustain in a highly competitive
market. Thus, a doubt whether these mergers are beneficial for getting technological aspects or it is
merely used to expand businesses.

(Phan et al., 2022) examined the performance of Indonesian banks concerning the impact of financial
technology. A total of 41 banks in Indonesia were taken as a sample to conduct the study. After running a
panel data regression model accompanying OLS and GMM tools, a negative impact was found due to the
global economic crisis and banks’ respective features. They examine the functioning value of banks in
Indonesia and find that profit growth and financial strength revert as firm liquidity after the adoption
remains ineffective, explaining that an implementation is a good approach for illustrating strength and
decency while lessening the focus on performance.

A study, by (Yoon et al., 2016) shows the aspects and prospects of financial technologies in the banking
industry. FinTech provides modern technology which has the ability to create a paradigm change in the
way that clients lead their lives by improving the user experience and enabling quick, easy, 24/7 banking.
There is a paradox in the literature concerning the acceptance of FinTech services despite the benefits it is
projected to bring, although FinTech services are said to provide enormous assistance and advantages to
both customers and financial institutions. If this inconsistency isn't resolved in the first phase of FinTech

Page |
service dissemination, a situation may develop in which newly launched FinTech services are employed
just partially by customers, rather than fully.

(Leong et al., 2018) in a study shows that by adopting financial technologies administration and overhead
costs are significantly reduced. Different techniques as means of making the process digitalized are taken
by the banks such as time-saving digital means of payment mechanisms rather than manual procedure of
transactions by staff. People can transact while staying at home, pay for different types of utility
payments, and get banking services 24 hours a. The study then concludes that these alternative and
advanced means of banking services would make a revolution in the banking industry all over the world
and the firms would be more efficient in delivering services, customer satisfaction, and increment of
financial inclusion. Lastly, these changes would positively affect the operating efficiency and profitability
of the fintech banking institutions.

(Qasim and Abu-Shanab, 2016) explore how the performance of the banks is affected by the adoption and
implementation of technologies and financial innovations. The author tries to classify the innovations into
different categories based on their functions. Some are for banks’ internal operations, and some are only
for customer applications. The study highlighted that these financial technologies expand the ways of
connecting with customers and make performance efficient. Moreover, some aspects increase businesses,
ensure financial stability and improve data management. Furthermore, these technologies are used to
manage information by combining reduction in financial layout and risks associated with banking
activities. The challenges that arise from legal compliance are also reduced.

In a study, (Naifar, 2019) tries to assess the possible influences of financial technologies and ways of
digitalization on the performance of the Islami economy and Islami banking. It was documented that
these innovations increase the efficacy of the economic performance of Islamic banks incorporating
financial stability. It is believed that if these advanced technologies are used continuously, the whole
economy will be benefited in the long run. A favorable connection between these innovations and the
performance of medium size and comparatively smaller firms has been found. It is expected that if the
banks can adopt this digitalization at the preliminary stage, they are supposed to make a wider expansion
of activities using e-commerce. Lastly, the paper provides advice that fintech banks can only be beneficial
if all the internal resources are efficiently used and in place. Fintech is more promising in the countries
which are developed and have a large uplifted infrastructure.

Page |
Chapter: 3 Empirical Methodology

The study's methodology is presented here. Here, data collection, sampling, and the usage of gathered
data will be discussed for analysis. This will provide a clear rationale for the study's scope and the
strategy for achieving its objectives. The very first of this methodology will talk about the research
strategy and then will show an in-depth overview of data collection and sampling, after that, this part will
talk about the variables and their implications. Finally, this report will talk about the limitation of this
dissertation.

3.1 Sample and Data

To get answers to the research questions, an empirical model has been developed. There is a total of three
types of data collected for conducting the analysis. They are cross-sectional data, time series data, and
panel data. The ten renowned private commercial banks are taken as samples for the analysis covering 10
years of data from 2012 to 2021. Panel Data is used to study the effects of technology on the performance
of the sample banks in Bangladesh while considering individual variation.

10 private commercial banks which are enlisted in Dhaka Stock Exchange are taken as samples over 16
variables for 10 years. The model incorporates a total of 100 observations of 10 samples for 16 variables.
The ten-sample private commercial banks are:

1. Dutch Bangla Bank Limited. 6. Southeast Bank Limited.

2. BRAC Bank Limited. 7. Trust Bank Limited.
3. Eastern Bank Limited. 8. International Finance Investment and Commerce
4. Prime Bank Limited. Bank Limited (IFIC Bank).
5. Mercantile Bank Limited. 9. Bank Asia Limited.
10. One Bank Limited.

3.2 Variables
Dependent variables: Four performance indicators are to be taken as dependent variables or explained
variables. The variables are Return on Asset (ROA), Return on Equity (ROE), Net Interest Margin (NIM),
and Basic Earning Power (BEP). ROA indicates how efficiently banks utilize their asset portfolio in
generating profits. ROE demonstrates the efficient use of banks’ common equity capital. The after-tax net
profit of the sample banks is to be considered as the return to calculate the ROA and ROE. Besides, the
use of ROA and ROE, net profit margin (NIM) plays a crucial role in indicating the performance of banks
(Siddik et al., 2016). Higher NIM dictates higher profitability of the banks. The Basic Earning

Page |
Power (BEP) gauges how well a firm generates earnings concerning its assets. Earnings Before Interest
and Taxes (EBIT) is divided by Total Assets in the BEP calculation.

Independent Variables: This paper aims to find out the implications of financial technology on the
banks’ overall performance. There is a total of 12 independent variables that have been taken to conduct
the analysis and for showing the impact on the dependent variables. The independent variables are;

1. Mobile Banking 7. Credit Risk

2. Online Banking 8. Capital Adequacy Ratio
3. Internet Payment Gateway 9. Firm Size
4. Agent Banking 10. GDP Growth
5. ATM 11. Inflation
6. Capital Ratio 12. Interest Rate Spread

Mobile banking, Online banking, Internet Payment Gateway, Agent banking, and ATM are qualitative
variables and so, are used as dummies here. These variables take values of either 0 or 1. When the value
is 1, it means the particular bank adopts these breakthroughs in that particular year, and if not, the value is
0. Capital Ratio, Credit Risk, Capital Adequacy Ratio, and Firm Size are taken as firm-specific factors
and work as control variables. To add the impacts of macroeconomic phenomena, GDP growth, Inflation,
and Interest Rate Spread have been taken for this model.

3.3 Research Model

This research will be conducted using the Ordinary Least Square (OLS) Model for the panel data
regression model. The model is given as follows.

PERFit = α1 + β2 MACROt + β3 Xit + β4 EBANKit + εi

PERF is the dependent variable, which is the performance of the bank as shown by ROA, ROE, NIM, and
BEP which are called proxy variables. Α1 is the model's intercept coefficient. Inflation and GDP growth
are market-specific variables, so they are in MACRO. Other control variables are in X. EBANK is the
"dummy" variable, and ε is the model's error term. This research model will be run using the ordinary
least square (OLS) Method, Fixed Effect Method, Random Effect, and Generalized Least Square(GLS)
Method.

The Model can be rewritten elaborately as

Page |
 Yit = α1 + β2 Xit + +β3Dit +εi

Where,
 Y= ROA, ROE, NIM, and BEP
 X6 = Inflation
 α = Intercept co-efficient/ Constant
 X7 = Interest Rate spread.
 β= Co-efficient of all the independent
 D1 = Mobile Banking
variables
 D2 = Online Banking
 X1 = Capital Ratio
 D3 = Internet Payment Gateway
 X2= Credit Risk
 D4 = ATM
 X3 = Capital Adequacy Ratio
 D5 = Agent Banking
 X4= Firm Size
 ε = Error Term
 X5 = GDP Growth

Table 1: A list of the variables and associated definitions

Variables Description Abbreviation Expectation

Return on Assets Net profit after tax/ Total Asset ROA +

Net profit after tax/ shareholders’
Return on Equity ROE +
Equity
Net interest income/ Earning
Net Interest Margin NIM +
Asset
Basic Earning Power Operating Income/ Total Asset BEP +
Capital Ratio Equity/ Total Asset CAP +
Credit Risk NPL/ Total Loans CR -
Firm Size Log of Total Asset FS ?
Regulatory capital/ Total risk-
Capital Adequacy Ratio CAR ?
weighted Assets
GDP Growth Percentage of GDP growth GDP +
Inflation Inflation Rate Inf +
The interest rate on a loan – the
Interest Rate Spread IRS +
interest rate on a deposit

Mobile banking, Online

Dummy variables. The value is 1 Mob, Online,
banking, Interest, Payment
if these variables are present in a IPG, ATM, +
Gateway, ATM Agent
particular year or 0 otherwise. Agent
banking
Source: Author’s Self Contribution

Page |
3.4 Limitations of the Study
The major weakness of the paper that is to be conducted, is mostly depending on secondary data instead
of primary data which would make the research more unparalleled, absolute, and supreme. Another
limitation of the paper is not the incorporation of all the banks operating in Bangladesh and the analytical
period is only 10 years. Moreover, due to the lack of availability of information, some other relevant
independent variables are not taken.

Page |
Chapter 4: Data Analysis

In this chapter, the data obtained from both resources will be organized and broken down into digestible
parts. The data will be analyzed before the findings are reached. In this part, quantitative tools will be
used to examine the quantitative data, and subjective tools like charts and graphs will be used to interpret
the quantitative data. In light of the findings from the literature review, the outcomes of the study and data
analyses will be illustrated in this chapter.

4.1 Regression Model Analysis

A regression model has been run using the Ordinary Least Square (OLS), Fixed Effect Model, Random
Effect, and Generalized Least Squares (GLS) method to analyze the impact of financial technologies and
associated control variables on the overall performance of the ten sample banks. The performance of the
banks is defined as including Return on Asset (ROA), Return on Equity (ROE), Net Interest Margin
(NIM), and Basic Earning Power as the indicators. Thus, four regression has been conducted on these
four dependent variables. The results of the regression model and associated interpretations are given as
follows;

4.1.1 ROA as a Dependent Variable

Table 2 demonstrates the picturesque statistics of the dependent and independent variables of ten private
commercial banks of Bangladesh for 2012-2021 using the four models of OLS, fe, re, and GLS. The
coefficients of the independent variables dictate the individual influences of these variables on the
explained variable Return on Asset. With a positive coefficient, we may infer that a rise in the
independent variable also results in an increase in the dependent variable's mean. The coefficient is
negative if there is a general downward trend in the dependent variable as the independent variable grows.
The coefficient of capital ratio is -.00518 which means if the credit ratio increases by 1 %, the ROA
would be decreased by .00518% holding the impact of another independent variable constant. The
Probability value of this coefficient is greater than 5% which indicates this explanatory variable does not
have a significant impact on the performance indicator variable. The negative coefficient variables are
credit risk, bank size, online banking, ATM, inflation, and interest rate spread. These coefficients indicate
that if these independent variables rise by one percentage, the ROA would be lessened by .12325, .0799,
00427, .0129, .8323, and .0041 percent respectively in the OLS model.

Page |
Table 2: Output of model coefficients, incorporating (Pooled Ordinary least square (OLS), Fixed effect
(fe), Random effect (re), Generalized Least square (GLS))

Estimates
Dependdent Variable:
Return on Asset (ROA) Ordianry Least Generalised Least
Fixed Effect (fe) Random Effect (re)
Square (ols) Square (gls)
cap -0.00157575 -0.00575806 -0.00157575 -0.00157575
cr -0.1232468 -0.05967575 -0.1232468 -.1232468*
bs -.07989891*** -.10355436*** -.07989891*** -.07989891***
car 0.12426089 0.07817077 0.12426089 0.12426089
mob .01661981*** .0148218*** .01661981*** .01661981***
on -0.00426726 -0.00069282 -0.00426726 -0.00426726
Explanatory
ipg 0.00527766 .00847215* 0.00527766 0.00527766
Variables
ag 0.00413385 0.00342608 0.00413385 0.00413385
atm -.01293133* -.01073036* -.01293133* -.01293133**
gdp 0.04558025 0.05498218 0.04558025 0.04558025
inf -.83234329* -1.494487*** -.83234329* -.83234329**
is -0.00406298 -.00862401** -0.00406298 -0.00406298
_cons .9647677*** 1.2849638*** .9647677*** .9647677***

N 100 100 100 100

R-sq 0.7368 0.7324
F 20.296398 27.414155
rho 0.39508928 0
chi2 243.55678 279.95032
sigma_e 0.00882309 0.00882309
sigma_u 0.00713054 0
Source: Author’s self-contribution based on output developed by Stata 12.0
Note: A level of significance of 5%, 1%, and 0.1% is indicated by the letters *, **, and ***.

Out of all the variables, bank size, and mobile banking are significant at a 0.1% significance level, ATM
is significant at a 5% significance level, and Inflation has a significance level of 5% (OLS), 0.1%(Fe),
and, 1%(GLS. Whereas the rest independent variables have insignificant influences as the P value is
greater than 5%. The variable of inflation got the highest standard error of 33% whereas the interest rate
spread secured the lowest one. As the probability value of the F test is 0.000, it is exhibited that all the
explanatory variables have a combined significant impact on the performance dictated by the ROA of the
banks.
Furthermore, The .7368 and .7324 R square values of the OLS and fixed effect model dictate that 74%
variability of the dependent variable ROA is explained by the dependent and control variables. This R
square value is an acceptable level and the higher this value, the higher the accuracy level of the model.
The adjusted R square is the R square adjusted by the explanatory variables. The value of the adjusted R

Page |
square is .7005 which I less than the R square value which means still other independent variables can be
added to the model that will increase the variation level and accuracy.
Rho, also known as the term inter-correlation coefficient, was calculated to be 0.3950 by using the fixed
effect method. That figure suggests that panel-to-panel differences account for 39.50 percent of the
overall variation in ROA. At the 5% level of significance, the estimated chi-square values for the random
effect method and the generalized least square (GLS) method are 243.68 and 249.33, respectively,
demonstrating the cumulative significance including all firm characteristics and systematic aspects
included in the model in explaining the non-performing loan of banks. Both the random effect method
and the generalized least squares (GLS) technique were used to get estimates for these variables.

Model Specificaton Test

This section describes the various model specification tests used to construct the model that assesses the
impact of Bank Specific variables and macroeconomic factors on the profitability of the ten commercial
banks in Bangladesh as measured by ROA.
Table 3: Output of Hausman Test for ROA

Hausman chi test (ROA)

Prob>chi2 0

Source: Author’s self-contribution based on output developed by Stata 12.0

Interpretation: The outcomes of the Hausman test determine whether the method, fixed effects or
random effects, will be used (described below). The alternative hypothesis asserts that random effects are
preferred over fixed ones, despite the fact that the null hypothesis predicts the opposite. We may conclude
that now the Random-effect model has been preferable to the Fixed-effect model as the output of Chi-
square is significant statistically at the 5% significance level (67.55).
Table 4: Output of B/P LM Test for ROA

Lagrangian multiplier
test (ROA)

Prob>chi2 1.00000

Source: Author’s self-contribution based on output developed by Stata 12.0

Interpretation: To conclude, the LM test that suggests picking between a Random effect and Pooled
OLS regression model is sound, given that we have considered that the variation between estimates is
zero. Thus, it may be concluded that the units under examination are comparable to each other (i.e. no
panel effect). Since we cannot establish that the panels are statistically different from one another (i.e.,

Page |
reject the null hypothesis), we must instead establish that pooled OLS or cross-sectional FGLS should
yield more reliable estimates than random-effect mode. Chi-square = 0.000, which is a statistically
insignificant value.

After evaluating the Pooled OLS technique against the Fixed effect method and the random effect method
using the Hausman test and the B/E LM test, we concluded that it was the most appropriate. To this end,
we shall use the OLS method in our diagnostic analysis.

Diagnostic Checks
Parametric diagnostics are used to assess the parameter estimates and look for data that have a significant,
unjustified impact on the analysis. Econometric tests are techniques for assessing how well a linear
regression that has been adapted to data captures the data's structure. Regression model diagnostics are
instruments that gauge a model's adherence to its underlying assumptions and look into it. Diagnostic
problems have been tested to detect whether the model contains any multicollinearity problems, omitted
variable bias, heteroscedasticity problems, and autocorrelation problems. The reason to testify these tests
is to ensure the model's accuracy. A diagnostic test's main function is to categorize or forecast the
existence or lack of a disorder. A diagnostic test's clinical performance is determined by how well it can
accurately divide subjects into pertinent subgroups. As additional tests are developed, it's critical to assess
how well they classify cases when compared to other tests or the standard.

Multicollinearity Test
Table 5: Multicollinearity Test for ROA
Variable VIF 1/VIF

inf 4.71 0.2124

ag 4.21 0.2375
is 4.02 0.2488
bs 2.57 0.3884
atm 2.43 0.411476
car 1.91 0.523353
on 1.84 0.544642
mob 1.82 0.549181
cap 1.64 0.611213
GDP 1.45 0.688491
cr 1.44 0.692938
ipg 1.2 0.833821

Mean VIF 2.44

Page |
Source: Author’s self-contribution based on output developed by Stata 12.0

Interpretation: Multicollinearity is the term used to describe significant correlation coefficients between
two or more separate variables in multiple regression modeling. Multicollinearity may result in biased or
deceitful conclusions whenever a researcher or analyst tries to assess how comprehensively particular
descriptive significant aspects can be used to predict or perceive the entirely reliant variable in
probabilistic reasoning (Bayman, 2021). The table above shows the regressed VIF values for our
independent variables, and they show that all of them, including capital ratio, credit risk, CAR, GDP,
CAR, inflation, mobile, and online banking do not have multicollinearity problems because their VIF
value is less than 5. Our model's mean VIF is 2.44, which shows that multicollinearity is not a problem
across the board for the entire model.

Heteroscedasticity Test
Table 6: Heteroscedasticity Test for ROA
chi2(1) = 237.6700
Prob > chi2 = 0.0000
Source: Author’s self-contribution based on output developed by Stata 12.0

Interpretation: Heteroskedasticity or heteroscedasticity occurs when the degrees of separation of a

forecasted variable are not consistent when seen over a wide range of separate variable readings or at
earlier time intervals. Heteroskedasticity is a term used to describe a scenario where the residual term or
standard error of a regression model exhibits a high degree of variability (Tovohery, 2020). If the error
variance is non-constant from one year to another year that means a heteroscedasticity problem exists
there. This is one of the tests of diagnostic checks of a model. It is a problem that a model suffers. It is
also called unequal variance. Whereas if the data set has equal variances, there is no heteroscedastic issue,
according to the null hypothesis, H0, depending on the heteroscedasticity test. It is stated that the set of
data has non-constant error variance, which causes a heteroscedasticity problem, whenever the set of data
has rejected the hypothesis h0. The chi-square p-value for the model is 0.000, which is less than 0.05. As
a result, the null hypothesis, H0, of equal variances is rejected as the heteroskedasticity test standard,
indicating that the model has a heteroscedastic issue. The bank's variance is not steady over time,
indicating that there has been a dramatic change in the variance.

Wooldridge Test of Autocorrelation

Table 7: Wooldridge Test of Autocorrelation for ROA
F (1,9) = 38.26
Prob > F = 0.0002
Source: Author’s self-contribution based on output developed by Stata 12.0

Page |
Interpretation: Autocorrelation is the degree to which the pertinent factors are associated during two
succeeding periods. It analyzes the link between a parameter's lagged version and its actual amount in
time-series data (Chand, 2018). The conventional linear regression method makes the assumption that
although data are gathered across time, the values of the disturbance factor are momentarily independent.
However, whenever the assumption is flawed, autocorrelation becomes a problem. The Wooldridge test
establishes the presence of autocorrelation in the framework when the H0, in which there is no 1st
autocorrelation, is rejected and the p-value is smaller than 0.05%. The F ratio and p-value for the model's
Wooldridge test are 38.26 and 0.0002, respectively. According to the conventional value, the hypothesis
is rejected, in this case, indicating that the model has autocorrelation. The independence of the variables is
shown by this

RAMSEY Reset Test

Table 8: RAMSEY Reset Test for ROA
F (3, 84) = 213.55
Prob>F = 0.000
Source: Author’s self-contribution based on output developed by Stata 12.

Interpretation: One such test for the linear regression model is the Ramsey Regression Equation
Specification Error Test (RESET). In particular, it checks whether the fitted values may be combined in
non-linear ways to better describe the response variable. The notion behind of test is that if the response
variable can be explained by a combination of explanatory variables that is not linear, then the model is
incorrectly specified, and a polynomial or some other quasi-functional form would be a better
approximation of the process that generated the data. It means dropping the relevant variable from the
model. If this happened then estimators, error variance, and standard of error are biased. As a result, our
significant test is also biased or insufficient. According to Ramsey RESET Test, the P value (.000) is less
than 5%. So, we should reject the null hypothesis (Ho) and conclude that there is omitted variable bias.
4.1.2 ROE as a Dependent variable

Table 7 demonstrates the picturesque statistics of the dependent and independent variables of ten private
commercial banks of Bangladesh for 2012-2021. The coefficients of the independent variables dictate the
individual influences of these variables on the explained variable Return on Equity. The coefficient of
capital ratio is -.10026 which means if the credit ratio increases by 1 %, the ROA would be decreased
by .10026 % holding the impact of other independent variables constant. The Probability value of this
coefficient is less than 5% which indicates this explanatory variable has a significant impact on the
performance indicator variable. The negative coefficient variables are credit risk, CAR, internet payment

Page |
gateway, agent banking, ATM, and interest rate spread. These coefficients indicate that if these
independent variables rise by one percentage, the ROA would be lessened by
1.25, .8411 .0039, .0204, .02302, and .00606 percent respectively.

Table 9: Output of model coefficients, incorporating (Pooled Ordinary least square (OLS), Fixed effect
(fe), Random effect (re), and Generalized Least square (GLS)) for ROE

Estimates
Dependdent Variable: Return
Generalised
on Equity (ROE) Ordianry Least
Fixed Effect (fe) Random Effect (re) Least Square
Square (ols)
(gls)
cap -.10025545** -.064385* .10025545*** -.10025545***
cr -1.2570898** -0.80581607 1.2570898*** -1.2570898***
bs 0.0317121 0.00789106 0.0317121 0.0317121
car -0.84117818 -1.1664144* 0.84117818 -.84117818*
mob 0.01903771 0.01829943 0.01903771 0.01903771
on 0.01011491 0.01482507 0.01011491 0.01011491
Explanatory
ipg -0.00389877 -0.00920611 0.00389877 -0.00389877
Variables
ag -0.0204007 -0.0380335 -0.0408014
atm -0.02301941 -0.02263937 0.02301941 -0.02301941
gdp 0.60563958 0.1771879 0.60563958 0.60563958
inf 0.71195563 -0.8670305 0.71195563 0.71195563
is -0.0060598 -0.01657572 -0.0121196
_cons -0.10821056 0.33992025 0.10821056 -0.10821056

N 100 100 100 100

R-sq 0.2808 0.2343
F 2.8295397 2.9775546
rho 0.27973352 0
chi2 33.954476 39.028134
sigma_e 0.0524479 0.0524479
sigma_u 0.0326854 0
Source: Author’s self-contribution based on output developed by Stata 12.0
Note: A level of significance of 5%, 1%, and 0.1% is indicated by the letters *, **, and ***.
The independent variables of the bank size, mobile banking, online banking, GDP growth, and inflation
have positive coefficients which mean if these variables increase by 1%, the explanatory variable ROA
would be magnified by 0.031712, .019038, .010115, 0.60564, and, 0.711956% respectively. Out of all the
variables, the capital ratio has a 0.1% significance level according to GLS and the fixed effect model and
credit has also the same level. variable of inflation got the highest standard error whereas the interest rate
spread secured the lowest one. As the probability value of the F test is 0.000, it is exhibited that all the
explanatory variables have a combined significant impact on the performance dictated by the ROE.
Furthermore, The .2808 and .2324 R square values of the OLS and fixed effect model dictate that 29%
variability of the dependent variable ROE is explained by the dependent and control variables. This R

Page |
square value is an acceptable level and the higher this value, the higher the accuracy level of the model.
The adjusted R square is the R square adjusted by the explanatory variables. The value of the adjusted R
square I is less than the R square value which means still other independent variables can be added to the
model that will increase the variation level and accuracy.
Rho, also known as the term inter-correlation coefficient, was calculated to be 0.2797 by using the fixed
effect method. That figure suggests that panel-to-panel differences account for 27.97% of the overall
variation in ROE. At the 5% level of significance, the estimated chi-square values for the random effect
method and the generalized least square (GLS) method are 33.95 and 39.028, respectively, demonstrating
the cumulative significance including all firm characteristics and systematic aspects included in the model
in explaining the non-performing loan of banks. Both the random effect method and the generalized least
squares (GLS) technique were used to get estimates for these variables.
Model Specification Test
This section describes the various model specification tests used to construct the model that assesses the
impact of Bank Specific variables and macroeconomic factors on the profitability of the ten commercial
banks in Bangladesh as measured by ROE.

Table 10: Output of Hausman Test for ROE

Hausman chi test
(ROE)
Prob>chi2 0.017

Source: Author’s self-contribution based on output developed by Stata 12

Interpretation: The outcomes of the Hausman test determine whether the method, fixed effects or
random effects, will be used (described below). The alternative hypothesis asserts that random effects are
preferred over fixed ones, even though the null hypothesis predicts the opposite. We may conclude that
now the Random-effect model has been preferable to the Fixed-effect model as the output of Chi-square
is significant statistically at the 5% significance level.

Table 11: Output of B/P LM Test for ROE

Lagrangian multiplier
test (ROE)

Prob>chi2 1.00000

Source: Author’s self-contribution based on output developed by Stata 12

Page |
Interpretation: To conclude, the LM test that suggests picking between a Random effect and Pooled
OLS regression model is sound, given that we have considered that the variation between estimates is
zero. Thus, it may be concluded that the units under examination are comparable to each other (i.e. no
panel effect). Since we cannot establish that the panels are statistically different from one another (i.e.,
reject the null hypothesis), we must instead establish that pooled OLS or cross-sectional FGLS should
yield more reliable estimates than random-effect mode. Chi-square = 0.000, which is a statistically
insignificant value, and thus, diagnostic checks will be concluded using OLS.

Multicollinearity Test
Table 12 Multicollinearity Test ROE
Variable VIF 1/VIF
inf 4.71 0.212416
ag 4.21 0.237514
is 4.02 0.248817
bs 2.57 0.388441
atm 2.43 0.411476
car 1.91 0.523353
on 1.84 0.544642
mob 1.82 0.549181
cap 1.64 0.611213
GDP 1.45 0.688491
cr 1.44 0.692938
ipg 1.2 0.833821
Mean VIF 2.44
Source: Author’s self-contribution based on output developed by Stata 12.0

Interpretation: The table above shows the regressed VIF values for our independent variables, and they
show that all of them, including capital ratio, credit risk, CAR, GDP, CAR, inflation, mobile, and online
banking do not have multicollinearity problems because their VIF value is less than 5. Our model's mean
VIF is 2.44, which shows that multicollinearity is not a problem across the board for the entire model.

Heteroscedasticity Test

Table 13: Heteroscedasticity Test for ROE

chi2(1) = 11.84
Prob > chi2 =
0.0006
Source: Author’s self-contribution based on output developed by Stata 12.0

Page |
Interpretation: The chi-square p-value for the model is 0.0006, which is less than 0.05. As a result, the
null hypothesis, H0, of equal variances is rejected as the heteroskedasticity test standard, indicating that
the model has a heteroscedastic issue. The bank's variance is not steady over time, indicating that there
has been a dramatic change in the variance

Wooldridge Test for Autocorrelation

Table 14: Wooldridge Test for Autocorrelation for ROE

F( 1, 9) = 11.308
Prob > F= 0.0084
Source: Author’s self-contribution based on output developed by Stata 12.0

Interpretation: The F ratio and p-value for the model's Wooldridge test are 11.308 and .0084,
respectively. According to the conventional value, the hypothesis is rejected, in this case, indicating that
the model has autocorrelation. The independence of the variables is shown by this.

RAMSEY Reset Test

Table 15: RAMSEY Reset Test for ROE

F(3, 84) = 0.91

Prob > F = 0.4390

Source: Author’s self-contribution based on output developed by Stata 12.0

Interpretation: The notion behind of test is that if the response variable can be explained by a
combination of explanatory variables that is not linear, then the model is incorrectly specified, and a
polynomial or some other quasi-functional form would be a better approximation of the process that
generated the data means dropping the relevant variable from the model. If this happened then estimators,
error variance, and standard of error are biased. As a result, our significant test is also biased or
insufficient. According to Ramsey RESET Test, the P value (.4390) is greater than 5%. So, we should
accept the null hypothesis (Ho) and conclude that there is no omitted variable bias

Page |
4.1.3 NIM As A Dependent variable
Table 16: Output of model coefficients, incorporating (Pooled Ordinary least square (OLS), Fixed effect
(fe), Random effect (re), and Generalized Least square (GLS)) for ROE.

Estimates
Dependdent Variable: Net
Profit Margin (NIM) Ordianry Least Generalised Least
Fixed Effect (fe) Random Effect (re)
Square (ols) Square (gls)
cap 0.06804494 0.03965385 0.06804494 0.06804494
cr 0.17854427 0.78739052 0.17854427 0.17854427
bs .18756516* .20064063* .18756516* .18756516**
car 1.135414 2.3196136* 1.135414 1.135414
mob .10275398* .14847252** .10275398* .10275398*
on -.15702856* -.22754184*** -.15702856* -.15702856**
Explanatory
ipg 0.05579208 0.00005258 0.05579208 0.05579208
Variables
ag 0.00988891 0.02957148 0.00988891 0.00988891
atm -.14703888** -.12142355* -.14703888** -.14703888**
gdp -0.42558472 -0.32415494 -0.42558472 -0.42558472
inf 8.6320396* 10.688264** 8.6320396* 8.6320396**
is .06903984* .08678921** .06903984* .06903984*
_cons -2.8241765** -3.290109** -2.8241765** -2.8241765**

N 100 100 100 100

R-sq 0.3086 0.2904
F 3.2357121 3.9138014
rho 0.29734094 0
chi2 38.828546 44.630512
sigma_e 0.09926381 0.09926381
sigma_u 0.06457226 0
Source: Author’s self-contribution based on output developed by Stata 12.0
Note: A level of significance of 5%, 1%, and 0.1% is indicated by the letters *, **, and ***.

Page |
Interpretation: The above table demonstrates the picturesque statistics of the dependent and independent
variables of ten private commercial banks of Bangladesh for 2012-2021. The coefficients of the
independent variables dictate the individual influences of these variables on the explained variable Net
Interest Margin (NIM. The coefficient of capital ratio is .069045 which means if the credit ratio increases
by 1%, the NIM would be increased by .069045% holding the impact of another independent variable
constant. The Probability value of this coefficient is greater than 5% which indicates this explanatory
variable does not have a significant impact on the performance indicator variable. The negative
coefficient variables are online banking, ATMs, and GDP. These coefficients indicate that if these
independent variables rise by one percentage, the NIM would be lessened by 0.15703, 0.14704, and -
0.42558 percent respectively for OLS. The independent variables of the credit risk, bank size, capital
adequacy ratio, mobile banking, internet payment gateway, agent banking, GDP growth, and inflation
have positive coefficients which mean if these variables increase by 1%, the explanatory variable NIM
would be magnified by their respective values. Out of all the variables, bank size and, mobile banking is
significant at a 5% level, online banking is also significant at 0.1% significant (FE), and ATM, Inflation,
and interest rate spread have a significant level of 1%. As the probability value of the F test is 0.0007, it is
exhibited that all the explanatory variables have a combined significant impact on the performance
dictated by the NIM of the banks.

Furthermore, The .3086 and .2904 R square values of the OLS and fixed effect model dictate that a 30%
variability of the dependent variable ROE is explained by the dependent and control variables. This R
square value is an acceptable level and the higher this value, the higher the accuracy level of the model.
The adjusted R square is the R square adjusted by the explanatory variables.

Rho, also known as the term inter-correlation coefficient, was calculated to be 0.2973 by using the fixed
effect method. That figure suggests that panel-to-panel differences account for 29.73 percent of the
overall variation in NIM. At the 5% level of significance, the estimated chi-square values for the random
effect method and the generalized least square (GLS) method are 38.82 and 44.43, respectively,
demonstrating the cumulative significance including all firm characteristics and systematic aspects
included in the model in explaining the non-performing loan of banks. Both the random effect method
and the generalized least squares (GLS) technique were used to get estimates for these variables.

Model Specification Test

This section describes the various model specification tests used to construct the model that assesses the
impact of Bank Specific variables and macroeconomic factors on the profitability of the ten commercial
banks in Bangladesh as measured by NIM.

Page |
Table 17: Output of Hausman Test for NIM

Hausman chi test (NIM)

Prob>chi2 0.04

Source: Author’s self-contribution based on output developed by Stata 12.0

Interpretation: The outcomes of the Hausman test determine whether the method, fixed effects or
random effects, will be used. (The alternative hypothesis asserts that random effects are preferred over
fixed ones, even though the null hypothesis predicts the opposite. We may conclude that now the
Random-effect model has been preferable to the Fixed-effect model as the output of Chi-square is
significant statistically at the 5% significance level.

Table 18: Output of B/P LM Test for NIM

Lagrangian multiplier
test (NIM)

Prob>chi2 1.00000

Source: Author’s self-contribution based on output developed by Stata 12.0

Interpretation: To conclude, the LM test that suggests picking between a Random effect and Pooled
OLS regression model is sound, given that we have considered that the variation between estimates is
zero. Thus, it may be concluded that the units under examination are comparable to each other (i.e. no
panel effect). Since we cannot establish that the panels are statistically different from one another (i.e.,
reject the null hypothesis), we must instead establish that pooled OLS or cross-sectional FGLS should
yield more reliable estimates than random-effect mode. Chi-square = 0.000, which is a statistically
insignificant value, and thus, diagnostic checks will be concluded using OLS.

Multicollinearity Test
Table 19:Multicollinearity Test for NIM
Variable VIF 1/VIF

inf 4.71 0.212416

ag 4.21 0.237514
is 4.02 0.248817
bs 2.57 0.388441
atm 2.43 0.411476

Page |
 car 1.91 0.523353
on 1.84 0.544642
mob 1.82 0.549181
cap 1.64 0.611213
GDP 1.45 0.688491
cr 1.44 0.692938
ipg 1.2 0.833821

Mean VIF 2.44

Source: Author’s self-contribution based on output developed by Stata 12.0

Interpretation: The table above shows the regressed VIF values for our independent variables, and they
show that all of them, including capital ratio, credit risk, CAR, GDP, CAR, inflation, mobile, and online
banking do not have multicollinearity problems because their VIF value is less than 5. Our model's mean
VIF is 2.44, which shows that multicollinearity is not a problem across the board for the entire model.

Heteroscedasticity Test
Table 20: Heteroscedasticity Test for NIM
chi2(1) = 632.64
Prob > chi2 = 0.0000
Source: Author’s self-contribution based on output developed by Stata 12.0

Interpretation: The chi-square p-value for the model is 0.0006, which is less than 0.05. As a result, the
null hypothesis, H0, of equal variances is rejected as the heteroskedasticity test standard, indicating that
the model has a heteroscedastic issue. The bank's variance is not steady over time, indicating that there
has been a dramatic change in the variance

Wooldridge Test for Autocorrelation

Table 21: Wooldridge Test of Autocorrelation for NIM

F( 1, 9) = 0.402
Prob > F = 0.5417

Source: Author’s self-contribution based on output developed by Stata 12.0

Interpretation: The F ratio and p-value for the model's Wooldridge test are 0.402 and 0.5417,
respectively. According to the conventional value, the hypothesis is accepted, in this case, indicating that
the model has no autocorrelation problem. The independence of the variables is shown by this.
RAMSEY Reset Test

Page |
Table 22: RAMSEY Reset Test for NIM
F(3, 84) = 230.34
Prob > F = 0.0000
Source: Author’s self-contribution based on output developed by Stata 12.0

Interpretation: The notion behind of test is that if the response variable can be explained by a
combination of explanatory variables that is not linear, then the model is incorrectly specified, and a
polynomial or some other quasi-functional form would be a better approximation of the process that
generated the data. It means dropping the relevant variable from the model. If this happened then
estimators, error variance, and standard of error are biased. As a result, our significant test is also biased
or insufficient. According to Ramsey RESET Test, the P value (0.000) is less than 5%. So, we should
reject the null hypothesis (Ho) and conclude that there is omitted variable bias.

4.1.4 BEP As A Dependent Variable

Table 23: Output of model coefficients, incorporating (Pooled Ordinary least square (OLS), Fixed effect
(fe), Random effect (re), and Generalized Least square (GLS)) for BEP.

Estimates
Dependdent Variable: Basic
Generalised
Power Earning (BEP) Ordianry Least Random Effect
Fixed Effect (fe) Least Square
Square (ols) (re)
(gls)
cap 0.00441685 -0.01036445 0.00441685 0.00441685
cr -0.11409056 -0.02640176 -0.11409056 -0.11409056
bs -.22123509*** -.27871352*** -.22123509*** -.22123509***
car 0.39520346 0.24526505 .39520346* .39520346*
mob .04357795*** .038888*** .04357795*** .04357795***
on -0.01978257 -0.00901189 -0.01978257 -0.01978257
Explanatory
ipg 0.01540126 .02452201* 0.01540126 0.01540126
Variables
ag 0.01394947 0.01383292 0.01394947 0.01394947
atm -.03192737* -.02651173* -.03192737* -.03192737**
gdp 0.05121198 0.12917155 0.05121198 0.05121198
inf -2.4318405** -4.0200157*** -2.4318405** -2.4318405**
is -0.01059845 -.02183253** -0.01059845 -0.01059845
_cons 2.6636539*** 3.4427705*** 2.6636539*** 2.6636539***

N 100 100 100 100

R-sq 0.7555 0.7521
F 22.415712 31.074833
rho 0.42170546 0
chi2 268.98855 309.18224
sigma_e 0.02195507 0.02195507
sigma_u 0.01874843 0

Page |
Source: Author’s self-contribution based on output developed by Stata 12.0
Note: A level of significance of 5%, 1%, and 0.1% is indicated by the letters *, **, and ***

Interpretation: The above table demonstrates the picturesque statistics of the dependent and independent
variables of ten private commercial banks of Bangladesh for 2012-2021. The coefficients of the
independent variables dictate the individual influences of these variables on the explained variable basic
earning power. The coefficient of capital ratio is .00441 which means if the credit ratio increases by 1 %,
the BEP would be increased by .00441% holding the impact of other independent variables constant. The
Probability value of this coefficient is greater than 5% which indicates this explanatory variable does not
have a significant impact on the performance indicator variable. The negative coefficient variables are
credit risk, bank size, online banking, ATM, inflation, and interest rate spread. These coefficients indicate
that if these independent variables rise by one percentage, the BEP would be lessened by their respective
percentage. Out of all the variables, bank size, mobile banking, and Inflation have a significance level of
0.1%. Whereas ATMs and, CAR is significant at 5%, and inflation is at 1%. As the probability value of
the F test is 0.000, it is exhibited that all the explanatory variables have a combined significant impact on
the performance dictated by the BEP of the banks.

Furthermore, The .7552 and .7521 R square values of the OLS and fixed effect model dictate that a 75%
variability of the dependent variable ROE is explained by the dependent and control variables. This R
square value is an acceptable level and the higher this value, the higher the accuracy level of the model.
The adjusted R square is the R square adjusted by the explanatory variables.

Rho, also known as the term inter-correlation coefficient, was calculated to be 0.4217 by using the fixed
effect method. That figure suggests that panel-to-panel differences account for 42.17 percent of the
overall variation in BEP. At the 5% level of significance, the estimated chi-square values for the random
effect method and the generalized least square (GLS) method are 268.98 and 309.18, respectively,
demonstrating the cumulative significance including all firm characteristics and systematic aspects
included in the model in explaining the non-performing loan of banks. Both the random effect method
and the generalized least squares (GLS) technique were used to get estimates for these variables

Model Specification Test

This section describes the various model specification tests used to construct the model that assesses the
impact of Bank Specific variables and macroeconomic factors on the profitability of the ten commercial
banks in Bangladesh as measured by BEP.

Table 24: Output of Hausman Test for BEP

Page |
 Hausman chi test
(BEP)
Prob>chi2 0.03

Source: Author’s self-contribution based on output developed by Stata 12.0

Interpretation: The outcomes of the Hausman test determine whether the method, fixed effects or
random effects, will be used. (The alternative hypothesis asserts that random effects are preferred over
fixed ones, even though the null hypothesis predicts the opposite. We may conclude that now the
Random-effect model has been preferable to the Fixed-effect model as the output of Chi-square is
significant statistically at the 5% significance level.

Table 25: Output of B/P LM Test for BEP

Lagrangian multiplier
test (BEP)

Prob>chi2 1.00000

Source: Author’s self-contribution based on output developed by Stata 12.0

Interpretation: To conclude, the LM test that suggests picking between a Random effect and Pooled
OLS regression model is sound, given that we have considered that the variation between estimates is
zero. Thus, it may be concluded that the units under examination are comparable to each other (i.e. no
panel effect). Since we cannot establish that the panels are statistically different from one another (i.e.,
reject the null hypothesis), we must instead establish that pooled OLS or cross-sectional FGLS should
yield more reliable estimates than random-effect mode. Chi-square = 0.000, which is a statistically
insignificant value, and thus, diagnostic checks will be concluded using OLS.

Multicollinearity Test
Table 26: Multicollinearity Test for BEP

Page |
Variable VIF 1/VIF

inf 4.71 0.212473

ag 4.2 0.238134
is 3.8 0.263001
bs 2.57 0.389529
atm 2.42 0.412587
car 1.9 0.527623
on 1.83 0.546359
mob 1.82 0.55086
cap 1.37 0.729575
cr 1.33 0.750571
ipg 1.2 0.834
Mean VIF 2.47

Interpretation: The table above shows the regressed VIF values for our independent variables, and they
show that all of them, including capital ratio, credit risk, CAR, GDP, CAR, inflation, mobile, and online
banking do not have multicollinearity problems because their VIF value is less than 5. Our model's mean
VIF is 2.44, which shows that multicollinearity is not a problem across the board for the entire model.

Heteroscedasticity Test
Table 27: Heteroscedasticity Test for BEP
chi2(1) = 245.54
Prob > chi2 = 0.0000
Source: Author’s self-contribution based on output developed by Stata 12.0

Interpretation: The chi-square p-value for the model is 0.0006, which is less than 0.05. As a result, the
null hypothesis, H0, of equal variances is rejected as the heteroskedasticity test standard, indicating that
the model has a heteroscedastic issue. The bank's variance is not steady over time, indicating that there
has been a dramatic change in the variance.

Wooldridge Test for Autocorrelation

Table 28: Wooldridge Test of Autocorrelation for BEP

F( 1, 9) = 75.128
Prob > F = 0.0000

Source: Author’s self-contribution based on output developed by Stata 12.0

Page |
Interpretation: The F ratio and p-value for the model's Wooldridge test are 75.128 and .000,
respectively. According to the conventional value, the hypothesis is rejected, in this case, indicating that
the model has autocorrelation. The independence of the variables is shown by this

RAMSEY Reset Test

Table 29: RAMSEY Reset Test for BEP
F(3, 85) = 503.17
Prob > F = 0.0000
Source: Author’s self-contribution based on output developed by Stata 12.0

Interpretation: The notion behind of test is that if the response variable can be explained by a
combination of explanatory variables that is not linear, then the model is incorrectly specified, and a
polynomial or some other quasi-functional form would be a better approximation of the process that
generated the data. If this happened then estimators, error variance, and standard of error are biased. As a
result, our significant test is also biased or insufficient. According to Ramsey RESET Test, the P value
(0.000) is l ess than 5%. So, we should reject the null hypothesis (Ho) and conclude that there is omitted
variable bias.

4.2 Trend Analysis

4.2.1 Numbers of Agents

No. of Agents
1200000
1000000
800000
600000
400000
200000
0
2017 2018 2019 2020 2021

No. of agents

Figure 1: Numbers of Agents of the Sample Banks

Interpretation: The graph displays the analysis of the data of the agent banking counted on the taken
sample of ten commercial banks in Bangladesh. Here, a 2017–2021 trend analysis is displayed. The trend

Page |
indicates a significant and positive correlation between the agents and the year. The trend is rising at an
accelerating rate.

4.2.2 Numbers of Registered Customers

No. of registered accounts (in Lac)

1400
1200
1213.9
1000
993.36
800
796.49
600 675.2
588
400
200
0
2017 2018 2019 2020 2021

Figure 2: No. of MFS Industry registered accounts

Interpretation: Here, the statistics display an examination of the year's registered consumers' positive
trends from 2017 to 2021. The trend is growing at an accelerating rate and is upward-sloping. The number
of clients at this location is growing rapidly. The rate is 9.93 crores from 2019 to 2020, an increase of
about 1 crore from the previous year as a result of Covid-19 as well as other digital reforms in
Bangladesh.

4.2.3 Numbers of Effective Account

No. of Operative Acoount in (LAC)

2021 312.52
2020 323.27
2019 347.63
2018 373.13
2017 210
25 75 125 175 225 275 325 375
2017 2018 2019 2020 2021
No. of Effective accounts 210 373.13 347.63 323.27 312.52

No. of Effective accounts Linear (No. of Effective accounts )

Figure 3: No. of Effective accounts of MFS

Page |
Interpretation: Data on the number of registered subscribers reveals a substantial upward trend between
2017 and 2021. 2018 is the year with the most active accounts on this site. However , there was a
progressive decline in the number of active customers in the MFS sector in 2019 and 2020. The
consequences of the Covid-19 outbreak in Bangladesh might be the cause.

4.2.4 Amount of All Transactions

Total transactions in tk (crore)

65000
57556.88
45000 41647.64
29571 33105.57
25000 22213.67
5000
2017 2018 2019 2020 2021
Total transaction in tk 22213.67 29571 33105.57 41647.64 57556.88
(crore)

Total transaction in tk (crore)

Figure 4: Total Amount of Transactions of MFS Industry

Interpretation: The graph above analyzes the aggregate volume of transactions made through the MFS
in a given year. The entire transaction for the year is displayed here in crores. The year 2020 will see the
most transactions, and the pattern has a positive trend. The volume of transactions is rapidly increasing.

4.2.5 Total Number of ATMs

Numbers of ATMs
14000
12000
10000
8000
6000
4000
2000
0
2017 2018 2019 2020 2021

numbers of atms

Figure 5: Numbers of ATMs

Page |
Interpretation: The graph displays the total number of ATMs counted on the taken sample of ten
commercial banks in Bangladesh. Here, a 2017–2021 trend analysis is displayed. This 5-year trend
indicates a significant and upward-sloping pattern that means the number of ATMs and their uses is
increasing rapidly. The trend is rising at an accelerating rate.

4.3 Comparative Analysis

4.3.1 Areas of service comparison

shares
0% 0% 1% 0.002, 2%

30% 48%

18%

Brac( Bkash) DBBL (Rocket) Nagad Mercantie (mycash)

Trust bank(T cash) UCB ( Ucash) Souhteast (Telecash) others

Figure 5: Areas of service comparison

Interpretation: The service coverage area has been analyzed based on the mobile financial service
provided taking data from the ten sample banks also including Nagad as a competitor. The MFS
Industry's share distribution is seen in this image. The largest part belongs to bKash (48%) Inaugurated by
Brac Bank Ltd.. the second, is expanding quickly in Nagad. Around 18% share has been captured by
Rocket launched by DBBL. Here, bKash, Nagad, and Rocket control around 95% of the company, with
the remaining 5%.

Page |
4.3.2 Total Subscribers

Subscribers in million
Subscribers in million

92.69
54
32.256
Brac( Bkas DBBL Nagad 2.2
Mercantie 2
Trust 0.95
UCB 0.027
( U- Souhteast 0.5
others
h) (Rocket) (mycash) bank(T cash) (Telecash)
cash)
Subscribers 92.69 54 32.256 2.2 2 0.95 0.027 0.5
in million

Figure 6: Total Subscribers

Interpretation: Based on the mobile financial service offered and data from the 10 sample banks, as well
as Nagad as a competitor, the total users or subscribers in million, have been examined. Here, from the
graph, it has been exhibited that Bkash has the highest market customer subscribers, where DBBL got the
second position and southeast has the lowest number.
4.3.3 ATMs

ATMs
32
Souhteast (Telecash)
152
162
Trust bank(T cash)
103
450
DBBL (Rocket) 4907

Brac( Bkash) 375

0 1000 2000 3000 4000 5000 6000

ATMs

Figure 7: Number of ATMs

Interpretation: The number of ATMs owned by the sample banks is shown in this data graphic. Due to
DBBL's excellent brand image for branch banking and ATMs, the Rocket has the most ATMs of any
company in this area. Mercantile bank which has 450 ATMs, is in second place, while bKash is in third
place.

Page |
4.3.4 Agents

Agent number

Ok mobile banking 90000

20000
UCB ( Ucash) 95274
116000
Trust bank(T cash) 65000
239003
DBBL (Rocket) 240000
200000
Brac (Bkash) 300000
0 50000 100000 150000 200000 250000 300000 350000

agent number

Figure :8 Number of Agents

Interpretation: This data chart illustrates the agent number of sample banks. Here, southeast bank
(Telecash) has the fewest agents and BRAC (Bkash) has the most. UCB and MBL have a sizable number
of agents despite their relative youth on the market. DBBL is the third agent in terms of number.
4.3.5 Branches

NO. of Branches

220
187
146 136
129 133
113 107
85

Brac DBBL Mercantile Trust one bank Prime IFIC EBL Southeast
Bank

Branches

Figure 12: Number of Branches

Interpretation: The number of branches of the ten banks has been highlighted in the above graph to
assess the market capture ability of the banks. From the graph, it is highlighted that DBBL is the bank that
has the highest position in spreading its business indicated by the largest number of branches. The second
position is secured by BRAC bank and EBL has the lowest number of branches.
4.3.6 ROA, ROE, and NIM Comparison

Page |
 17%
15% 14% 14%
11% 12%
10%
8% 8%
7%
4% 5% 5%
3%
2% 2% 2% 2% 2%
1% 1% 1% 1% 1% 1% 1% 1% 1% 1%
Brac DBBL Mercan- Trust one0%
bank Prime IFIC EBL South- Bank
tile Bank east Asia
NIM 4% 3% 7% 2% 2% 5% 2% 2% 1% 2%
ROA 1% 1% 1% 1% 0% 1% 1% 1% 1% 1%
ROE 8% 15% 14% 14% 5% 11% 8% 17% 12% 10%

Figure 10: ROA, ROE, and NIM Comparison

Interpretation: The above graph shows the ROA, ROE, and NIM ratio of sample ten renowned private
commercial banks for the year 2021. In terms of ROA, all the banks got a similar percentage of values
except one bank which secured the lowest value. According to the ROE, EBL has the highest percentage
of 17%. DBBL is the second having 15% of the ratio, and Mercantile and trust have secured the third
position having 14% ROE. One bank again got the lowest percentage of 5%. From the aspects of Net
Interest Margin (NIM), Mercantile bank has the highest ratio of 7% and southeast has the lowest
percentage.

From the above analysis of data and trends, it is explicit that all the independent and control variables
have as a whole significant impact on the performance of sample ten commercial banks although the
individual significance level has been varied for these explained variables. From the diagnostic aspects,
there are no multicollinearity problems whereas other problems vary for the dependent variables. Trend
analysis exhibits that continuously the number of e-banking services and numbers of transactions is rising
at a growing speed. Thus, financial inclusion is also rising. Comparing different aspects of the sample
banks, it can be concluded Brac bank is far ahead in capturing the customer base through the largest
mobile financial service platform Bkash, whereas DBBL is creating competition in the industry rapidly
spreading its business through the highest number of branches, and ATMs. In terms of bank performance,
Mercantile bank has a wholesome position among all the banks.

Page |
Chapter 5: Findings
5.1 Findings of Regression Result

For all the dependent variables of ROA, ROE, NIM, and BEP the regression model shows that mobile
banking has a significant and positive relationship with the dependent variable which means the banks
can magnify their net profit in terms of total assets and total equity, operating profits in terms of total
asset, and also net interest profit to earning asset. Therefore, it may be assumed that in the year after
Mobile banking adoption, the banks started to make money after recouping their technology expenditures.
Additionally, Hernando and Nieto (2007) discover a favorable correlation between e-banking and an
organization's profitability with a one-and-a-half-year time lag. Except for ROE, for all the other three
variables, online banking has a negative impact. There was a consistent negative effect of e-banking on
NIM, even if it was not large in the year of adoption and the year after. Additionally, the present study
notices a new negative correlation between the use of e-banking systems and banks' profitability in the
following year of e-banking adoption. The regression findings indicate a substantial negative effect of
online banking on the margin of net interest among the four proxies for bank performance. As a
consequence, the study's findings show that the banks were unable to generate a satisfactory return on
their investments in the deployment of an e-banking system. With regard to each of the four proxy
variables, the analysis also predicts a favorable correlation between CAR, inflation, and bank
profitability. The OLS estimate reveals that default risk (Dr) has a favorable connection with NIM but a
substantial negative influence on ROA. Additionally, there is a strong and favorable correlation between
NIM and the loan-to-assets ratio (LOANS). This implies that banks may generate greater interest revenue
if they make a lot of advances and loans to clients. However, the study's findings also indicate that a large
percentage of non-performing loans will lower the overall return. As a result, bank management should
focus on improving the sound loan profile and lowering the number of problematic loans. According to
this analysis, there is an inverse link with bank size and profit, indicating that big banks are probably
making less money. In addition to the financial factors, the variables linked to Internet Payment Gateway,
ATMs, Agent Banking and GDP also demonstrate a significant correlation with banks' profitability. In
terms of ROA and NIM, the panel data estimate reveals a significant positive correlation between Agent
banking and bank profitability. This suggests that the banks’ agents are essential for safeguarding a bank's
profits. According to the data, banks' returns are adversely impacted by GDP growth, however, their
profitability is positively correlated for two variables with the inflation rate. The results indicate that
Bangladesh's commercial banks have failed to reap the rewards of innovations, and as a consequence,
their attempts to implement computerized banking systems have yielded negative returns.

Page |
5.2 Findings of Trend Analysis

The MFS sector and the concept of online banking are expanding quickly, having successfully overtaken
a decade to include about 77% of the population in financial inclusion. BB's rigorous rules and high
admission requirements prevented many banks from offering MFS services. However, The number of
Agents and thus, banking services are spreading rapidly with Bank Asia having the largest number of
agents. The number of clients at this location is growing rapidly. The rate is 9.93 crores from 2019 to
2020, an increase of about 1 crore from the previous year as a result of Covid-19 as well as other digital
reforms in Bangladesh. , there was a progressive decline in the number of active customers in the MFS
sector in 2019 and 2020. The consequences of the Covid-19 outbreak in Bangladesh might be the cause.
In 2020, most transactions have been made through online and mobile banking platforms, and the pattern
has a positive trend. The volume of transactions is rapidly increasing. Trend analysis exhibits that
continuously the number of e-banking services and numbers of transactions is rising at a growing speed.
Thus, financial inclusion is also rising.

5.3 Findings of Comparative Analysis

BKash has the highest market share (48%) formally opened by Brac Bank Ltd. About 95% of the
corporation is under the ownership of Kash, Nagad, and Rocket. According to research, Southeast has the
fewest branches and agents to distribute the service, DBBL has the second-largest number of market
customer subscribers, and bkash has the most due to its marketability and ease. The Rocket now has 4907
ATMs, making it the firm in this region with the most ATMs thanks to DBBL's stellar reputation for
branch banking and ATMs. By contrasting various characteristics of the sample banks, it can be said that
Brac Bank is far ahead in attracting customers through the biggest mobile financial service platform,
Bkash, while DBBL is quickly establishing competition in the sector by rapidly expanding its business
through the most branches and ATMs. Mercantile Bank is in a strong position among all the banks in
terms of performance.

Page |
Chapter 6: Conclusion
6.1 Conclusion

New technology is developing daily in the world of information technology. Compared to conventional
banking, technology is faster and cheaper. A flurry of queries about its conception, acceptability, and
possible hazards are also emerging in people's minds in tandem with it. If the current state of global
banking is anything to go by, Bangladeshi banks will need to act quickly to provide the groundwork for a
solid system of technology-driven banking if they want to remain competitive. Online banking, POS,
kiosks, mobile banking, ATMs, and other forms of electronic commerce are all fast gaining popularity.
New guidelines for the protection of client data have been given by Bangladesh Bank. Furthermore,
internet branch banking is growing in popularity in our nation. Though the future of technology in our
nation is promising, our study shows that just the availability of mobile banking and ATMs has a
favorable influence on our basic earning capacity. Despite the negative effects of online banking, we may
still hold out hope that they are the result of widespread ignorance. It could have some beneficial benefits
if it is adopted by all banking sector customers. Financial technology and the process of digitalization
have a significant impact on profitability, operating efficiency, and the area of competition although
because of high inauguration costs and inefficiency in creating economies of scale, adoption of
technology can not make an expected profit and give a negative impact. However, At the end of
December 2021, there were 123.82 million active internet users, while at the end of October 2022, there
were 181.43 Million active mobile phone customers. It is clear that technology is progressively affecting
every kind of person. As a result, it might be a potent weapon for commercial banks to draw in and keep
clients. Keeping up in the increasingly competitive banking industry of the present requires, among other
things, providing consumers with the latest in customer service technologies and more streamlined
service delivery options. Banks with a greater capacity for rapid technological adoption are rewarded
more handsomely than those with slower adoption rates.

6.2 Prospects of Fintech Banking in Bangladesh

The Bangladesh Railway is the proud owner of a wide variations network of optical fiber (1,800 km) that
basically relates to the railway course and connects the country's major cities and towns. The backbone of
Bangladesh's e-banking infrastructure may be found in this fiber optic network. Grameen Phone or Ranks
ITT, two major mobile phone companies in Bangladesh, employ the fiber optic network to extend their
coverage area to rural regions. It's promising that certain Correlations between financial and PCBs have

Page |
been making use of this network of optical fibers for handling electronic payments, automated teller
machine transactions, and point-of-sale operations. The country's 389 upazilas and 17 development hubs
now have digital telephone exchanges. There is an active effort being made to bring the remaining
upazilas into the digital exchange network. Meanwhile, in 2006, Bangladesh connected to the global
underwater cable system, officially becoming part of the information superhighway. There are now 64
operational ISPs out of a maximum of 159 that have been linked to this system. Internet speeds are
modest, ranging from 32 to 56 kilobits per second (kbps) for digital display and 64 to 8 Mbps (Mbps) for
broadband. In this case, the present abilities of the ICT industry are anticipated to develop swiftly in
putting all upazilas within internet services, which will help to broaden the reach of e-banking across the
nation, as a portion of the government's goal of constructing digital Bangladesh. Banking institutions have
a computer density of 1.64 per 100 workers. The ratio of computers per employee is 45.34 for FCBs but
just 0.41 for NCBs. For specialized banks, the most likely outcome is 0.43, which is quite close to the
NCBs' scenario. But the percentage for private commercial banks is far higher, at 4.94. While just 8.18
percent of a bank as a whole lacks a LAN, many financial institutions' branches are not even connected to
the WAN. All international banks in our country now use some kind of online banking, and they've spent
a lot on automating banking services. Although they were the first to introduce electronic banking
systems to Bangladesh, now the vast majority of the country's private banks operate with the use of such
technology. Some banks in our nation provide ATM (Automated Teller Machine) services, while others
provide virtual and still others provide electronic cash transmission, debit cards, credit cards, and similar
services. In recent years, the opportunities for e-banking in Bangladesh have increased thanks to the
government's focus on creating a digital Bangladesh, establishing an ICT park, increasing the allotment
for evolving Infrastructure facilities, nullifying taxes on computer peripherals, and other measures, such
as the automated testing system of banking industry guided by the Bangladesh Bank and contestability
even amongst the scheduled commercial banks in enhancing customer services.

6.3 Recommendations
 The economy of Bangladesh is one of the fastest-growing in the world. Consequently, financial
technologies have a fantastic chance to seize by extending financial inclusions to the whole
community by expanding these concepts through campaigns and technology learning. MFS can
help Bangladesh expand its economy and prepare for the 4th Industrial Revolution.
 All banks should focus on the digital literacy of the users.
 Electronic New Employee Orientation, In order to reduce risks and guarantee smooth onboarding
procedures, banks may benefit from biometric authentication's ability to streamline the Do-Not-
Target-Know-List (Know Your Customer) process.

Page |
 When it comes to BB, strict restrictions are needed to cut down on fraud.
 Biometrics can be integrated into banking applications to ensure clients' safety while making in-
app purchases or logging into their accounts remotely.
 Biometric identifiers, such as fingerprint scanners, may be installed in ATMs by banks to ensure
that only valid cardholders can withdraw money and it will decline electronic robbery as well as
will include more people under these services with trust.
 To assist consumers in avoiding becoming victims of fraud and to speed up the government's
transition to a cashless society, financial service providers should offer digital literacy services.
 To maintain a positive and significant profit margin above the installment and operating costs of
these digital technologies, banks have to bring more prospective customers under these services
so that economics of scale and economics of scope can be achieved in making a good profit
margin.

Page |
References
1. Aduda, J., & Kingoo, N. (2012). The relationship between electronic banking and financial
performance among commercial banks in Kenya.
2. Arrawatia, R., Misra, A., & Dawar, V. (2015). Bank competition and efficiency: empirical evidence
from the Indian market. International Journal of Law and Management.
3. Bayman, E. O., 2021. Multicollinearity in Logistic Regression Models. Anesthesia &
Analgesia, 133(2), pp. 362-365.
4. Chapman, J., Garratt, R., Hendry, S., McCormack, A. and McMahon, W., 2017. Project Jasper: Are
distributed wholesale payment systems feasible yet. Financial System.
5. Chen, X., You, X., & Chang, V. (2021). FinTech and commercial banks’ performance in China: A
leap forward or survival of the fittest? Technological Forecasting and Social Change.
6. Chindudzi, M.G., Maradze, M.T.C. And Nyoni, M.T., The Impact Of Digital Banking On The
Performance Of Commercial Banks In Zimbabwe.
7. Gomber, P., Kauffman, R. J., Parker, C., & Weber, B. W. (2018). On the fintech revolution:
Interpreting the forces of innovation, disruption, and transformation in financial services. Journal of
Management Information Systems,
8. Gozman, D., Liebenau, J., & Mangan, J. (2018). The innovation mechanisms of fintech startups:
Insights from SWIFT's innotribe competition. Journal of Management Information Systems.
9. Haller, S., & Siedschlag, I. (2011). Determinants of ICT adoption: Evidence from firm-level data,
Applied Economics.
10. Hernando, I. and Nieto, M.J., 2007. Is the Internet delivery channel changing banks’ performance?
The case of Spanish banks. Journal of Banking & Finance.
11. Hossain, M. I. (2021). Effects of E-Banking Adoption on the Financial Performance of State-Owned
Commercial Banks in Bangladesh. Information Resources Management Journal.
12. Katsiampa, P., McGuinness, P. B., Serbera, J. P., & Zhao, K. (2022). The financial and prudential
performance of Chinese banks and Fintech lenders in the era of digitalization. Review of
Quantitative Finance and Accounting.
13. Kim, M., Zoo, H., Lee, H. and Kang, J., 2018. Mobile financial services, financial inclusion, and
development: A systematic review of academic literature. The Electronic Journal of Information
Systems in Developing Countries.
14. Lee, S., 2017. Evaluation of mobile application in user's perspective: case of P2P lending Apps in
FinTech industry. KSII Transactions on Internet and Information Systems (TIIS).
15. Leong, K. and Sung, A., 2018. FinTech (Financial Technology): what is it and how to use
technologies to create business value in a fintech way?.

Page |
16. Mallat, N., Rossi, M. and Tuunainen, V.K., 2018. Mobile banking services. Communications of the
ACM.
17. Md, A.B., Anton, A.K. and Sanjida, P., 2010. Adoption of e-banking in Bangladesh: An exploratory
study. African journal of business management, 4(13), pp.2718-2727.
18. Naifar, N. ed., 2019. Impact of Financial Technology (FinTech) on Islamic Finance and Financial
Stability. IGI Global.
19. Nayak, R., 2018. A conceptual study on digitalization of banking-issues and challenges in rural
India. Journal Homepag.
20. Onay, C., & Ozsoz, E. (2013). The impact of internet banking on brick and mortar branches: Turkey
Bank.
21. Phan, D. H. B., Narayan, P. K., Rahman, R. E., & Hutabarat, A. R. (2020). Do financial technology
firms influence bank performance? Pacific-Basin Finance Journal.
22. Qasim, H. and Abu-Shanab, E., 2016. Drivers of mobile payment acceptance: The impact of network
externalities. Information Systems Frontiers.
23. Sufian, F. and Habibullah, M.S., 2010. Bank-specific, industry-specific and macroeconomic
determinants of bank efficiency: Empirical evidence from the Thai banking sector. Margin.
24. Taher, S.A. and Tsuji, M., 2022. An Overview of FinTech in Bangladesh: Problems and
Prospects. FinTech Development for Financial Inclusiveness.
25. Tao, R., Su, C. W., Naqvi, B., & Rizvi, S. K. A. (2022). Can fintech development pave the way for a
transition towards a low-carbon economy: A global perspective. Technological Forecasting and
Social Change.
26. Thakor, A.V., 2020. Fintech and banking: What do we know? Journal of Financial Intermediation.
27. Tovohery, J. M., 2020. Application of Equality Test of Coefficients of Variation to the
Heteroskedasticity Test.
28. Vukadin, A., Lemoine, J.F. and Badot, O., 2016. Opportunities and risks of combining shopping
experience and artistic elements in the same store.
29. Wahab, A., Alam, T.M. and Raza, M.M., 2021, November. Usability Evaluation of FinTech Mobile
Applications: A Statistical Approach. In 2021 International Conference on Innovative Computing
(ICIC).
30. Wang, Y., Xiuping, S., & Zhang, Q. (2021). Can fintech improve the efficiency of commercial
banks? An analysis based on big data. Research in International Business and Finance.
31. Wu, G., & Yuan, H. (2021). The impact of fintech on the profitability of commercial banks in China.
32. Yoon, B., Kim, J., You, Y., & Kim, S. (2016). A study on user behaviors for consulting of FinTech
479.

Page |
33. Zhao, Q., Tsai, P.H. and Wang, J.L., 2019. Improving financial service innovation strategies for
enhancing china’s banking industry competitive advantage during the fintech revolution: A Hybrid
MCDM model.
34. Zveryakov, M., Kovalenko, V., Sheludko, S. and Sharah, E., 2019. FinTech sector and banking
business: competition or symbiosis.

Page |