You are on page 1of 75

JOMO KENYATTA UNIVERSITY OF AGRICULTURE AND TECHNOLOGY

TO ANALYSE THE RELATIONSHIP BETWEEN CREDIT RISK MANAGEMENT

INDICATORS, PERFORMANCE, AND PROFITABILITY OF COMMERCIAL

BANKS IN KENYA

A research proposal submitted in partial fulfilment of the requirements of the degree of

Bachelor of Science in Financial Engineering at the Jomo Kenyatta University of

Agriculture and Technology

SUPERVISOR: Dr. Victor Musau

(2021)
DECLARATION

We declare that this research is our original work and it has not been submitted anywhere for

a degree award. Acknowledgments have been made where we have used other people's

ideas. We take full responsibility for any shortcomings that may be found in this project.

Signature........................................Date.......................

AMOS MUTENDE SCM222-0722/2017

Signature........................................Date.......................

CLIFF MUCANGI SCM222-0346/2017

Signature........................................Date.......................

GRACE KINYUA SCM222-0336/2017

Signature........................................Date.......................

LILIAN KATUNYU SCM222-0364/2017

Signature........................................Date.......................

SOLOMON KARANJA SCM222-0398/2017

This research has been submitted for examination with my approval as a university

supervisor

Signature....................................Date.........................

Dr. Victor Musau

ii
ACKNOWLEDGEMENT

We would like to express our deep and sincere gratitude to the Almighty God for bringing us

this far. He has been gracious to us by giving us wisdom and understanding. Secondly, we

would like to thank our research supervisor, Dr. Victor Musau, for providing invaluable

guidance throughout this research. His dynamism, vision, sincerity, and motivation deeply

inspired us. It was a great privilege and honour to work and study under his guidance. Lastly,

we would like to thank our friends and family who supported us and offered deep insights

into the study.

iii
TABLE OF CONTENTS

DECLARATION......................................................................................................................ii

ACKNOWLEDGEMENT......................................................................................................iii

TABLE OF CONTENTS........................................................................................................iv

LIST OF ABBREVIATIONS...............................................................................................vii

ABSTRACT.............................................................................................................................ix

CHAPTER ONE.......................................................................................................................1

INTRODUCTION....................................................................................................................1

1.1 Background of Study.................................................................................................1

1.1.1 Credit Risk Management....................................................................................2

1.1.2 Financial Performance and Profitability of Commercial Banks.........................2

1.2 Statement of the Problem..........................................................................................3

1.3 Study Objectives.......................................................................................................4

1.3.1 Main Objective...................................................................................................4

1.3.2 Specific objectives..............................................................................................4

1.4 Justification of Study.................................................................................................4

CHAPTER TWO.....................................................................................................................5

LITERATURE REVIEW........................................................................................................5

2.1 Introduction...............................................................................................................5

2.2 Credit Risk................................................................................................................6

2.3 Credit Risk Management...........................................................................................7

2.4 Performance and Profitability...................................................................................9

iv
2.5 Conclusion and Summary.......................................................................................11

CHAPTER THREE...............................................................................................................12

RESEARCH METHODOLOGY.........................................................................................12

3.1 Introduction.............................................................................................................12

3.2 The Research Design...............................................................................................12

3.3 The population of Study..........................................................................................12

3.4 Data Collection........................................................................................................13

3.5 Data Analysis..........................................................................................................13

3.6 Panel Data...............................................................................................................16

3.6.1 Balanced and Unbalanced Panels.....................................................................16

3.6.2 Panel Data Models............................................................................................18

3.7 Properties of Panel Data Estimators........................................................................23

3.8 Choosing between Pooled OLS and Random effects: Breusch-Pagan Lagrange

Multiplier (LM) Test.....................................................................................................24

3.9 Choosing Between Fixed and Random Effects: Hausman Test..............................24

3.10 Diagnostic Tests....................................................................................................25

REFERENCES.......................................................................................................................27

APPENDICES........................................................................................................................32

Workplan..............................................................................................................................32

Budget Estimation.........................................................................................................33

v
LIST OF TABLES

Table 3.1: Variables, Description, and Formulas..................................................................24

Table 3.2: An example of a Panel Data Table........................................................................29

vi
LIST OF ABBREVIATIONS

ABSA Amalgamated Banks of South Africa

CAR Capital Adequacy Ratio

CBK Central Bank of Kenya

CR Credit Risk

DR Debt Ratio

DW Durbin Watson

EL Expected Loss

FE Fixed Effects

FGLS Feasible Generalised Least Squares

GLS Generalized Least Squares

LM Lagrange Method

LGD Loan Given Default

LSDV Least Squares Dummy Variable

NCBA National Cooperative Business Association

NPLR Non-Performing Loan Ratio

NSE Nairobi Stock Exchange

OLS Ordinary least squares

ROA Return On Assets

ROE Return On Equity

vii
RE Random Effects

UL Unexpected Loss

TL Total Loans

viii
ABSTRACT

Credit risk management in the banking sector is important not only because of the Financial Crisis

experienced in recent years but also due to its greater impact on bank’s financial performance,

growth and survival. Credit loans is one of the key sources of income of commercial banks,

therefore managing the risk related to credit greatly impacts the bank’s profitability. This study

analyses the relationship between credit risk management indicators, financial performance and

profitability of the Kenyan commercial banks. Data from 6 selected commercial banks listed in NSE

for the period 2016 to 2020 has been collected and analyzed using Random Effect model which was

the consistent model based on Hausman test. In the model specification, return on Equity(ROE) and

return on Assets (ROA) were used as banks’ profitability and performance indicators while capital

adequacy ratio (CAR), debt ratio (DR), bank size (BS), loan loss provision ratio (LLPR), non-

performing loan ratio (NPLR) were used as indicators of credit risk management.

The findings indicate that credit risk management has significant impact on the profitability and

performance of the Kenyan commercial banks. Results show that non-performing loans, capital

adequacy ratio, and bank loan loss provision have a negative impact on banks’ performance and

profitability. On the other hand, debt ratio, size of bank are found to have a negative impact

profitability while size of banks has a positive impact on performance of the commercial banks.

However, size of bank and debt ratio turned out to be not significant variables in explaining the

bank’s performance and profitability. The study thus recommends an effective credit risk

management for commercial banks of Kenya by maintaining an optimum level of capital adequacy

ratio, controlling and monitoring non-performing loan, ensure leverage to enhance financial

performance and profitability of the commercial banks.

ix
CHAPTER ONE

INTRODUCTION

1.1 Background of the Study

The financial sector in Kenya comprises insurance companies, commercial banks, pension

funds, and mortgages oriented historically to meet the needs of external trade and large-scale

commerce. Commercial banks are the most active players in the financial system and, in

particular, in the financial markets among the various financial institutions. Commercial

banks are regarded as financial institutions that primarily intermediate between the lenders

and borrowers. This means that they receive deposits from their clients who have a surplus

and lend the money to the clients with cash deficits (borrowers). They also provide capital for

the development of infrastructure and the creation of jobs. Therefore, commercial banks have

become an inherent part of society, in the industries and individual consumers.

Financial intermediation, however, may give rise to different types of risks with different

weights. Some of the risks include credit, liquidity, market, reputational, and operational

risks. Among the plethora of risks, the literature reveals that the commonly investigated risk

is credit risk. It’s the oldest form of risk in the capital market and is probably the main risk

for commercial banks. Anthony (1997) asserts that credit risk arises when non-performing

borrowers are unable or unwilling to perform in the pre-committed contract. Many

commercial banks have suffered losses due to the high non-performing loans ratio, which

indicates credit risk management. As a result, credit risk management in commercial banks

has become a crucial concept that influences the continuity, growth, and profitability of

commercial banks.

Commercial banks have varying regulatory policies which determine how well they mitigate

their risks. This, to a greater extent, affects the type of credit facilities they give to clients and

1
their investment decisions, thus affecting their performance and profitability. According to

Umoh (2002), not many banks can withstand a persistent run; even when a good lender of

last resort exists, depositors withdraw all their funds. The bank runs short of funds, hence

lacking liquidity support. At long last, the commercial bank forcefully closes its doors. This

means that the risks the banks are exposed to are endogenous while others are exogenous.

Although banks have been developing new policies to guide their daily activities and protect

themselves against such risks, they are still vulnerable to shocks mainly due to credit risks.

1.1.1 Credit Risk Management

Credit risk management is the practice of mitigating losses by understanding the adequacy of

a bank’s capital and loan loss reserves at any given time. This process has long been a

challenge for financial institutions. Credit risk management involves minimizing bank risk

and adjusting the risk rate of return by maintaining credit risk exposure. This is done with a

view of shielding the bank from the adverse effects of credit risk.

Basel Committee on Banking Supervision (1999) observed that banks increasingly face credit

risk (or counterparty risk) in various financial instruments other than loans, including

acceptances, interbank transactions, trade financing foreign exchange transactions, financial

futures, swaps, bonds, equities, and options. Since credit is one of the primary sources of

revenue-generating activity in commercial banks, credit risk management affects the

profitability and performance of banks. Moreover, poor management and poor control of

credit risk lead to excessive risk of a bank and impacts its profitability and performance.

1.1.2 Financial Performance and Profitability of Commercial Banks

The financial performance of a commercial bank refers to its ability to generate new

resources from daily operations over a given period. Several factors determine this within the

external and internal environment. Bobakovia (2003) explains that the profitability of any

2
The commercial bank highly depends on their ability to envisage, avoid and monitor risk.

Credit risk has a higher impact on commercial banks than the other risks since it has a direct

solvency threat to these financial institutions. The banks should therefore be able to

cover losses caused by uncertainties and risks as they occur.

The non-performing loans (NPLs) have a notable adverse effect on the banks’ performance

and profitability. Thus, poor risk management leads to poor financial performance hence the

closure of such commercial banks. In addition to the non-performing loans, the size of the

bank, debt ratio, capital adequacy ratio, loan loss provision also affect the performance and

profitability of banks and will be used in our study. Return on Assets (ROA) and the Return

on Equity (ROE) are the leading profitability indicators in banks.

1.2 Statement of the Problem

This research seeks to answer the question, “What is the impact of credit risk management

indicators on the performance and profitability of the commercial banks in Kenya since 2015-

2020?”. Previous studies have been conducted on the effects of credit risk management on

banks’ performance however, this research question has barely been covered by analysts

from the viewpoint of the Kenyan context in general, and in the context of the top-performing

commercial banks listed in Nairobi Securities Exchange. In particular, Kenya Commercial

Bank, Equity Bank, Cooperative Bank, Standard Chartered Bank, I&M, and DTB Bank will

serve as the sample commercial banks over the period 2015-2020. This study, therefore, aims

to investigate the viability of investing in credit risk management in commercial banks and

also to examine the impacts of the credit risk management determinants on the banks’

performance and profitability. This study will involve panel data analysis techniques mainly,

focusing on the fixed and random-effects models.

3
1.3 Study Objectives

1.3.1 Main Objective

To study the relationship between credit risk management indicators, performance, and
profitability of commercial banks in Kenya.

1.3.2 Specific Objectives

I. To model credit risk management indicators using a Fixed or Random-Effects model.


II. To determine the most efficient model between; Pooled OLS, Random Effects, and
Fixed Effects, using the Breusch-Pagan Langrage Multiplier Test and the Hausman
Test.
III. To test the significance of credit risk management indicators on banks' performance
and profitability using ANOVA.

1.4 Justification of the Study

Commercial banks are the most dominant in many world economies compared to other

financial institutions offering instalment loans. As a result, the financial intermediation role

banks play can’t be overlooked. They are vital for developing economies to achieve higher

economic growth. However, their role gives rise to different types of risks, mainly credit risk.

Therefore, this study is important in understanding how credit risk impacts the performance

of Kenyan banks, the type and strength of such impact, and the appropriate strategies that

increase banks’ efficiency in managing credit risk

4
CHAPTER TWO

LITERATURE

REVIEW

2.1 Introduction

This chapter presents the literature related to credit risk and credit risk management. It

includes how an increase in credit risk will raise the marginal cost of debt and equity, which

in turn increases the cost of funds for the bank.

Previous studies involving measuring the performance and profitability of banks have

traditionally used financial ratios; Return on Assets and Return on Equity as measures of

performance and profitability (Mathuva 2009; Wet & Toit 2006). The financial ratios have

proved to be useful in the interpretation of a company’s financial status and management

(Halkos & Salamouris, 2004). ( Wet &Toit,2006) asserts that ROE is one of the best

determinants of a bank’s performance since it combines the aspect of profitability, financial

leverage, and efficiency. Halkos and Salamouris (2004) also stress that ROE and ROA are

useful when making internal and external comparisons and predictions can easily be done.

The review of the related studies on the impact of credit risk on banks’ performance and

profitability has revealed mixed results, which are either positive or negative relationships

(Abbas et al., 2019). Ogboi & Unuafe (2013) analysed the impact of credit risk and capital

adequacy on the financial banks in Nigeria. Time series data obtained from the annual

financial reports of banks in Nigeria since 2004-2009 was used. They used the fixed-effect

model to determine the relationship between different credit risks which include (loans and

loan losses, capital adequacy, loans and advances, and bad loans) and ROA. The empirical

results pointed out that good management of credit risk and capital adequacy has a significant

5
positive impact on ROA while, loans and advances were reported to hurt the ROA of the

banks during the specified period.

2.2 Credit Risk

Basel Committee on Banking Supervision (1999) has defined credit risk as to the potential

that a bank borrower or counterparty will fail to meet its obligation by agreed terms.

Heffernan (1996), observed that credit risk is the risk that causes an asset or a loan to become

irrecoverable in the case of outright default, or the risk of delay in the servicing of the

loan. Bessis (2002), opined that Credit risk is critical since the default of a small number of

important customers can generate large losses, which can lead to insolvency. This risk is

determined by factors extraneous to the bank such as general unemployment levels, changing

socio-economic conditions, debtors’ attitudes, and political issues.

Kithinji (2010) analyzed the effect of credit risk measured by the ratio of loans and advances

on total assets and the ratio of non-performing loans to total loans and advances on return on

the total asset in Kenyan banks from 2004 to 2008. The study found that the bulk of the

profits of commercial banks are influenced by the amount of credit and non-performing

loans. The study provides the rationale to consider other variables that could impact on

bank’s performance and also a longer period of the study to capture the real picture of the

banks’ performance. Hence this study included the impact of liquidity and market risk as

components of the financial risk.

Afriyieet al. (2011) examined the impact of credit risk on the profitability of rural and

community banks in the BrongAhafo Region of Ghana. The study used the financial

statements of ten rural banks from the period of 2006 to 2010 (five years) for analysis. The

panel regression model was employed for the estimation. In the model, of Return on Equity

(ROE) and Return on Asset (ROA) were used as profitability indicators while Non-

Performing
6
Loans Ratio (NLPR) and Capital Adequacy Ratio (CAR) as credit risk management

indicators. The findings indicated a significant positive relationship between non-performing

loans and rural banks’ profitability revealing that there are higher loan losses but banks still

earn profit. He found that there is a relationship between credit risk management and

profitability of selected rural banks in Ghana.

Credit risk remains widely regarded as the major influence on a bank’s performance and the

major cause of bank failures, largely due to their limited capacity to absorb losses from bad

loans (AlMazrooei & Al-Tamimi,2007; Boffey & Robson,1995). According to Boffey and

Robson (1995), a bank’s capacity to absorb bad loans comes mainly from its profits and its

capital. A single substantial bad loan can have such a significant impact on the business.

Therefore, banks must manage their credit risks proactively. These losses are generally

categorized into three namely:

I. Expected loss (EL), which is classed as predictable and counted as part of the cost of

business thus is factored in the pricing.

II. Unexpected Loss (UL) which are unanticipated losses above the expected.

III. Loss Given Default (LGD), which refers to the loss incurred by the bank with loan

default.

2.3 Credit Risk Management

Athanasoglou et al. (2005) observe that the bank's role remains central in financing economic

activity, and its effectiveness could positively impact the overall economy. A sound and

profitable banking sector can withstand negative shocks and contribute to the financial

system's stability. Bank performance determinants have attracted the interest of academic

research as well as of bank management.

7
Studies dealing with internal determinants employ variables such as size, capital adequacy,

NPLR, LLPR, and DR. Poor asset quality and low levels of liquidity are the two major causes

of bank failures and are represented as the key risk sources in terms of credit and liquidity

risk. However, credit risk is by far the most significant risk faced by banks. The success of

their business depends on accurate measurement and efficient management of this risk to a

greater extent than any other risk (Gieseche, 2004). However, every bank needs to identify,

measure, monitor, and control credit risk to determine how credit risks could be lowered.

This means that a bank should hold adequate capital against these risks to compensate for the

risks incurred adequately.

The statistical evidence from the research conducted by Sparaford (1988) showed that

98 % of bank failures resulted from incidents related to poor asset quality due to poor loan

policies, non-compliance with policies and guidelines, and inadequate supervision. Sparaford

(1988) further asserts that the factors highlighted above are a result of poor credit culture, a

position confirmed by Colquitt (2007) and Boffey and Robson (1995). Expounding on this

concept, Colquitt (2007) proposes that a bank's credit culture determines the attitude, style,

perception, and behavior that will be exhibited by the bank and is primarily determined by the

attitude of management towards credit risk and could be in conflict with the policies of the

bank.

Credit risk management arises any time bank funds are extended, committed, invested, or

otherwise exposed through actual or implied contractual agreements, whether reflected on or

off the balance sheet. Literature review reveals that bank-specific or macroeconomic

variables are used as explanatory variables of the studies on the main determinants of credit

risk. Basel Committee on Banking Supervision (2006) observed that historical experience

8
shows that the concentration of credit risk in asset portfolios has been one of the major causes

of bank distress.

According to Robert and Gary (1994), the most apparent characteristic of failed banks is not

poor operating efficiency, however, but an increased volume of non-performing loans. Non-

performing loans in failed banks have typically been associated with regional macroeconomic

problems. DeYoung and Whalen (1994) observed that superior managers run their banks

cost-efficiently and thus generate large profits relative to their peers and impose better loan

underwriting and monitoring standards than peers, which results in better credit quality.

2.4 Performance and Profitability

Bobakovia (2003) asserts that a bank's profitability depends on its ability to foresee, avoid

and monitor risk, possibly to cover losses brought about by chance. The net effect of

increasing the ratio of substandard credits in the bank's credit portfolio decreases its

profitability. The bank's supervisors are well aware of this problem. However, it is

challenging to persuade bank managers to follow more prudent credit policies during an

economic upturn, especially in a highly competitive environment.

The conservative managers might find market pressure for higher profits tough to overcome.

For instance, Philip (1994) observed that the deregulation of the financial system in Nigeria

embarked upon from 1986 allowed the influx of banks into the banking industry. As a result

of alternative interest rates on deposits and loans, credits were given out indiscriminately

without proper credit appraisal. These inappropriate credit appraisal systems make banks

have non-performing loans that exceed 50 percent of the bank's loan portfolio.

Bourke (1989) reported that the effects of credit risk on profitability appear to be negative.

This result may be explained by taking into account that the more financial institutions are

exposed to high-risk loans, the higher the accumulation of unpaid loans. This implies that

9
these loan losses have produced lower returns to many commercial banks (Miller and Noulas,

1997; Kolapo et al., 2012). The findings of Felix and Claudine (2008) also show that return

on equity (ROE) and return on asset (ROA), which point to profitability, were negatively

related to the ratio of non-performing loans to total loans (NPL/TL) of financial institutions.

This, in turn, shows that there is a decrease in profitability.

However, the ratios are not oblivious of their shortcomings. Highlighting some of the

deficiencies, Oberholzer and Westhuizen (2004) and Chen and Yeh (1997) assert that these

ratios have limitations in their capacity to give a robust measurement of a bank's performance

and indeed the performance of firms in general. According to them, the ratios are inadequate

as measures of future performance since they are drawn from past performance. They further

emphasize that the ratios are measures of short-term performance. They lump together all the

aspects of the bank's performance, making it impossible to identify specific areas where

actual performance has been outstanding or below expectation. Thus, analysis drawn from

them can be seen as the starting point for any future research.

In addition, Lei (2005), while emphasizing that financial ratios remain a quick, helpful, and

reliable means of analyzing the performance of banks, acknowledges that the accuracy of

financial ratios may be distorted by inflation and the timing of the release of financial reports.

Other criticisms state that ratios ignore the importance of different parameters, such as the

cost of capital (Colquitt 2007). Others say that they are subject to manipulation within

acceptable accounting standards (Wet and Toit 2006). Consequently, several other

approaches have been employed to measure the all-around performance of banks, one of

which is the Data Envelopment Analysis. The Data Envelopment Analysis is being

researched, adopted, and compared to financial ratios (Ho and Zhu, 2004; Halkos and

Salamouris, 2004; Oberholzer and Westhuizen, 2004; Chen and Yeh, 1997).

10
2.5 Conclusion and Summary

Credit risk management is essential in optimizing financial institutions’ performance and

profitability, from almost all the literature reviewed. Additionally, effective credit risk

management practices involve establishing an appropriate credit risk environment, managers

operating a sound credit granting process, maintaining appropriate credit administration

involving the identification, analysis, and monitoring process, and adequate control over

credit risk.

Considering the empirical evidence, there is a need therefore to adopt the best credit risk

management practices that minimize losses arising from loans. This needs to be an active

process that enables banks to proactively manage their loan portfolios, minimize provisions

and lost interest income. Banks shall improve their performance and profitability through the

adoption of better credit risk management practices that involve methods, procedures,

processes, and rules used in minimizing loan losses facing them in their lending functions.

11
CHAPTER THREE

RESEARCH METHODOLOGY

3.1 Introduction

This chapter discusses the overall methodology of the study. It covers the research design, the

population of the study, data collection methods, and data analysis methods that we will

adopt for this study.

3.2 The Research Design

According to Saunders, Lewis & Thornhill (2011), descriptive research design involves

measuring a set of variables as they exist naturally. In this research project, we shall adopt

descriptive analysis. It helps answer immediate questions about the current state of affairs

(Mathews & Kostelis, 2011). A particular field of study may provide in-depth information

about the subject's characteristics under study (Houser, 2011). It emphasizes producing data

based on real-world observation through a purposeful and structured approach (Denscombe,

2003).

3.3 Population of the Study

The study population includes the people or items from which a study sample comes

(Zikmund et al., 2013). According to Mugenda and Mugenda (2003), It consists of the sets of

individuals or objects with similar observable characteristics distinct from the attributes of

other populations. In line with these definitions, the population of this study includes six

commercial banks in Kenya. The said banks are Kenya Commercial Bank, Equity Bank,

Cooperative Bank, Standard Chartered Bank, ABSA Bank, and Stanbic Bank.

12
3.4 Data Collection

The study will utilize secondary data contained in the CBK’s website and library. The audited

annual financial reports are reliable and readily available as all the commercial banks file

yearly performance reports with the CBK yearly.

3.5 Data Analysis

Data analysis systematically describes, illustrates, condenses, and evaluates data using

statistical and logical techniques. A variety of analytical procedures provide ways of drawing

inductive inferences from the data that has been collected by highlighting the phenomena of

interest from statistical fluctuations that may be in the data (Shamoo & Resnik,2003). This

research will utilize quarterly data for five years. The data will be presented in the form of a

panel and, R-program will be used to analyze the data by running pooled OLS, fixed, and

random-effects models. The best fit model will then analyze the relationship between the

credit risk management indicators with the banks’ performance and profitability.

This research paper will employ the panel data technique with the following econometric

model where the variables are explained in Table 3.1

𝑌𝑖𝑡 = 𝛽0 + 𝛽1𝑋1𝑖𝑡 + 𝛽2𝑋2𝑖𝑡 + 𝛽3𝑋3𝑖𝑡 + 𝛽4𝑋4𝑖𝑡 + 𝛽5𝑋5𝑖𝑡 + 𝜖𝑖𝑡 (3.1)

where: 𝛽0 represents the constant coefficient of variation.

Β𝑖 for 𝑖>0 shows the slope coefficient for the independent variables.

𝜖𝑖𝑡 refers to the error term

The subscript 𝑖 represents the index of the commercial bank.

The subscript t represents the quarterly time.

13
Table 3.1: Variables, Description, and Formulas.

Variables Description Formula


Y1= ROA Return on Total Assets Net Income/
Total Assets
Y2= ROE Return on Total Equity Net Income/
Shareholder Equity
X1= NPLR Non-performing Loans Ratio Non-performing loans/
Total loans
X2= CAR Credit Adequacy Ratio Total Capital/
Total Risk-Weighted Assets
X3= DR Debt Ratio Total debt/
Total Assets
X4= LLP Loan Loss Provision Ratio Loan Loss Provision/
Gross Loan
X5 =SIZE OF BANK Size of bank The logarithm of Total
Assets

Return on equity (ROE)- Return on equity (ROE) is a dependent variable and measures the

bank's return on shareholders' investment. ROE has been used as an indicator of profitability

in regression analysis because ROE and ROA have been used extensively in previous

research.

Return on assets (ROA)- Return on assets (ROA) is a dependent variable and measures the

bank's earnings before interest and taxes against its total net assets. It measures the efficiency

in managing banks to generate profits from their scarce resource. In addition, this ratio is

strongly influenced by the credit institution's provisioning policy since the net result

incorporates the cost of risk and the assets appear net of the provision in the bank balance

sheet. It is used to evaluate the performance of a credit institution (Sylvie de Goussergues et

al., 2010).

14
Non-performing loan ratio (NPLR)- Non-performing loans (NPL) are independent

variables that indicate how banks manage their credit risks. It defines the proportion of non-

performing losses to the total amount of the loan (Hosna et al., 2009). When the quantity of

this ratio increases, it will send a wrong message for management at the bank level because it

shows a high probability of not recovering funds granted to customers by banks. We expect

non-performing loans to have a negative relationship with profitability.

Capital adequacy ratio (CAR)- (CAR) is also an independent variable. It measures the basis

of a bank's financial strength from the point of view of a regulator. This ratio protects

depositors from unexpected losses. The capital adequacy ratio consists of the types of

financial capital considered the most reliable and the most liquid, mainly the bank's equity,

with a good sufficiency ratio.

Bank size- There exists a direct relation between a bank's size and profitability. When all

other factors are kept constant, the size of the banks determines the level of risks that its

clients are exposed to. A good-performing bank has more assets to keep it operating its

activities even in times of financial crisis. This means that loans offered by big banks are

most likely to be repaid compared to small banks.

Debt ratio (DR)-This measures the extent of a company's leverage. The debt ratio is defined

as total debt to total assets, expressed as a decimal or percentage. It's the proportion of a

company's assets that is financed by debt.

Loan Loss Provision Ratio (LLPR)- The loan loss provision ratio indicates how protected a

bank is against future losses. A higher ratio means the bank can withstand future losses

better, including unexpected losses beyond the loan loss provision.

15
3.6 Panel Data

This is a combined time-series cross-sectional dataset containing repeated observations on a

selection of variables from a set of observational units. A cross-sectional data set consists of

observations on a certain number of variables at a certain point in time, whereas a time-series

data set consists of a variable or several variables of observations over several periods.

3.6.1 Balanced and Unbalanced Panels

A panel data set consists of n sets of observations on individuals, which are usually denoted

as i = 1. . . N. If each individual in the panel data set is observed the same number of times,

usually denoted T, the data set is then referred to as a balanced panel. In our research, the

panel will be balanced, with N observations being equal to 6. An unbalanced panel data set,

on the other hand, is one in which individuals are observed different numbers of times. This

time is denoted as Ti.

16
Table 3.2: An example of a Panel Data Table

Dependent Independent variables


variables
Bank Year ROE ROA CAR NPLR DEBT SIZE LLPR
(quarterly RATIO OF
period) FIRM

KCB 2015
Q1 9.8 1.20 14.21 9.39 1.13 9.89 0.0048
Q2 6.34 0.75 13.77 8.50 1.35 9.98 0.0076
Q3 11.62 1.84 14.33 9.22 1.40 9.32 0.0023
Q4 11.94 1.44 14.73 10.05 1.26 9.45 0.0049
EQUITY 2015
Q1 5.60 1.50 11.47 7.98 1.76 12.87 0.0018
Q2 6.00 1.40 15.90 9.76 0.98 11.23 0.0023
Q3 9.61 0.63 17.45 9.90 0.76 10.50 0.0043
Q4 7.64 1.20 17.83 9.98 1.87 11.34 0.0054
ABSA 2015
Q1 9.76 1.11 13.05 8.90 1.76 8.89 0.0065
Q2 7.09 1.54 16.65 9.98 1.56 8.78 0.0017
Q3 10.60 1.17 20.54 11.00 1.43 9.02 0.0053
Q4 9.05 0.78 17.07 10.74 1.22 9.13 0.0032

There are three important types of variations in panel data;

Overall Variation- Variation over time and individuals.

Between variation- variation between individuals.

Within variation- variation within individuals.

1
Individual mean 𝑥̅ = ∑ ∑𝑥 (3.2)
𝑇 𝑖 𝑡 𝑖𝑡

17
𝟏
Overall mean ̅𝐱 = 𝚺 𝚺𝐱 (3.3)
𝐍𝐓 𝐢 𝐭 𝐢𝐭
𝟏
Overall variance 𝐒 𝟐 = 𝚺 𝚺 (𝐱 − 𝒙̅ )𝟐 (3.4)
𝐎 𝐍𝐓−𝟏 𝐢 𝐭 𝐢𝐭

𝟏
Between variance 𝐒 𝟐 = 𝚺 (𝐱̅ − 𝐱̅ )𝟐 (3.5)
𝐁 𝐍𝐓−𝟏 𝐢 𝐢

𝟏 𝟏
Within variance 𝐒 𝟐 = 𝚺 𝚺 (𝐱 − 𝐱̅ )𝟐 = 𝚺 𝚺 (𝐱 − 𝐱̅ + 𝐱̅ )𝟐 (3.6)
𝐖 𝐍𝐓−𝟏 𝐢 𝐭 𝐢𝐭 𝐢 𝐍𝐓−𝟏 𝐢 𝐭 𝐢𝐭 𝐢

The overall variation can be decomposed into between variation and within the variation

𝟐
𝐒𝐎 = 𝐒𝟐𝐁+ 𝐒𝟐 𝐖 (3.7)

We need to check the data to see if the between or within variation is larger for each variable.

3.6.2 Panel Data Models

Panel data models describe the individual behaviour both across time and across individuals.

There are three types of models;

3.6.2.1 The Pooled Model

In the panel data context, this is also called the population-averaged model. The pooled

model specifies constant on the coefficients α and β. The model is given by

𝐲𝐢𝐭 = 𝛂 + 𝐱β𝐢𝐭′ + 𝐮𝐢𝐭 (3.8)

It assumes that any form of heterogeneity in the model has averaged out during the cross-

sectional analysis. It applies OLS estimation ignoring the fact that it is panel data.

The OLS Estimator of β

A simple linear regression equation is given by:

Y = Xβ + ϵ (3.8.1)

18
Where: Y represents the vector of dependent/response variables.

Β represents the vector of regression parameters.

X represents the vector of explanatory/independent/regressor variables.

ϵ represents the vector of independent normal variables.

The OLS estimator of β is given as:

β̂ = (X′X)−1 X′Y (3.8.2)

However, the model cannot be applied in this research since the error terms are correlated.

This calls for other improved models, as discussed further below.

3.6.2.2 Individual-specific Effects Model

We assume that there is unobserved heterogeneity across individuals captured by αi.

The main question is whether the individual-specific effects αi are correlated with the regressors,

if they are correlated, a fixed-effects model is used. If they are not correlated, the random-effects

model is used.

The Fixed-effects Model. (FE)

A fixed-effect model examines individual differences in intercepts, assuming the same slopes

and constant variance across individuals (group and entity).

𝐲𝐢𝐭 = αi + 𝐱β𝐢𝐭′ + 𝐮𝐢𝐭 (3.9)

Since an individual-specific effect is time-invariant and is considered as part of the intercept,

αi is allowed to be correlated with other regressors. The model is estimated by within and

between effect estimations.

19
Within Estimation

Within estimation analyzes the variation within each individual or entity.

The “within” estimation is given by:

Yit−Yi=β1(xit−xi)+(uit−ui),for t=1,2,3…T

The left-hand side is called the time demeaned.

The OLS is applied on the time demeaned equation whereby the errors will not be correlated

and we will have cross-sectional error uj uk.

Between Estimation

The “between groups” estimation, also known as the group mean regression, analyses the

variation between individual entities (groups). To be specific, this estimation calculates group

means of the dependent and independent variables and thus reduces the number of

observations down to n. OLS is then performed on the transformed aggregated data: y = α +

x+ε

The FE model assumes that there is a correlation between the individual-specific effects αi

with the regressors x. We include αi as intercepts.

Each individual has a different intercept term and the same slope parameters.

𝐲𝐢𝐭 = αi + 𝐱β𝐢𝐭′ + 𝐮𝐢𝐭 (3.9.2)

We can recover the individual-specific effects after estimation as:

𝛂̂ 𝐢 = 𝐲̅𝐢 − 𝐱β𝐢𝐭 (3.9.3)

In other words, the individual-specific effects are the leftover variation in the dependent

variable that cannot be explained by the regressors.

20
21
3.6.2.3 Random Effects Model (RE)

A random-effects model assumes that individual effect (heterogeneity) is not correlated with

any regressor and estimates error variance specific to groups (or times). Thus, 𝛂𝐢 is an

individual-specific random heterogeneity or a component of the composite error term. This

explains why the random effect model is also called an error component model. The slopes

and intercepts of the regressors are the same across the individual. The difference among

individuals (or periods) lies in their specific errors, not in their intercepts. i.e.

The composite error term 𝛆𝐢𝐭 = 𝛂𝐢 + 𝐞𝐢𝐭

𝐲𝐢𝐭 = 𝐱′ 𝐢𝐭𝛃 + (𝛂𝐢 + 𝐞𝐢𝐭 )

Here 𝐯𝐚𝐫(𝐞𝐢𝐭 ) = 𝛔𝟐 + 𝛔𝟐 and 𝐜𝐨𝐯(𝛆 𝛂


𝛂 𝐞 𝐢𝐭 , 𝐢𝐬 ) = 𝛔𝟐
𝛆

𝛔𝟐𝛂
So 𝛒 = 𝐜𝐨𝐫(𝛆 𝛆 ) = (3.10)
𝛆 𝐢𝐭 𝐢𝐬 (𝛔𝛂𝟐 +𝐞𝛔𝟐 )

Rho is the fraction of the variance in the error due to the individual-specific effects. It

approaches 1 if the individual effects dominate the idiosyncratic error. A random-effect

model reduces the number of parameters to be estimated but produces inconsistent estimates

when an individual-specific random effect is correlated with regressors. (Greene, 2008: 200-

201).

The model is estimated by generalized least squares (GLS) when a covariance structure Σ

(sigma), is known.

22
Generalized Least Square Method

Introduce P to a linear equation Y=Xβ+ϵ, where P is an n×n matrix.

From this, the new linear equation is given by:

Y = PXβ + Pϵ

From the derivation of the estimate β using the OLS method, we get the estimate as shown

below:

β̂ = (X′X)−1 X′Y becomes (X′P′XP)−1 X′ PY = (X′P′PX)−1 X′P′PY

But P’P= Ω-1

Therefore, the new estimate

β̂ = (X′Ω−1 X)−1 X′Ω−1 Y (3.11)

Which is an efficient estimator of β and thus it is known as the GLS.

It should be known that the GLS estimators are unbiased if they have a zero-conditional

mean;

E[(ϵ|X)] = 0

Similarly, the variance of the estimator is given by:

Var(β̂) = δ2 (X′P′XP)−1 (3.11.1)

The GLS becomes OLS under homoscedasticity whenΩ−1 = I, where I represent the identity

matrix.

23
Limitation of GLS Method

The method is not sufficient since it cannot determine Ω or Σ. This leaves one with two

options; either to assume Σ or estimate it empirically. FGLS is the most efficient method in

solving this problem.

The Feasible Generalized Least Squares Method (FGLS)

The feasible generalized least squares (FGLS) is used to estimate the entire variance-

covariance matrix when Σ is not known. This is done by estimating Σ empirically. This

method replaces Ω in GLS to Ω̂ . Therefore, the FGLS estimator of β is given as:

β̂ = (X′Ω−
̂ 1 X)−1 X′Ω−
̂ 1Y (3.12)

Wooldridge provides a more flexible approach to estimating the FGLS;

Var(ϵ|X) = ϵ2 = δ2exp(δ0 + δ1x1 + δ2x2 + ⋯ + δnxn

By introducing the natural log on both sides, we get:

ln(ϵ̂2 ) = α0 + δ1 x1 + δ2 x2 + ⋯ δn xn + e

Assume that the predicted value from this model is given by

̂
ĝi = ln(ϵ2 )

This can be converted by introducing exponential into weights assigned wi such that:

̂
ŵi = exp(ĝi ) = exp (ln (ϵ2 )) = ϵ̂̂ (3.13)

3.7 Properties of Panel Data Estimators

We prefer estimators that are consistent and efficient. We check for consistency first then for

efficiency.

24
Consistency

The distribution of 𝛃̂𝐧 collapses on β as n becomes large:

𝐩𝐥𝐢𝐦 𝛃̂𝐧 = 𝛃

Consistency is established based on the law of large numbers.

If an estimator is consistent, more observations will tend to provide more precise and

accurate estimates.

Efficiency

Efficiency (minimum variance) is usually established relative to specific classes of estimators

Example: OLS is efficient (minimum variance) among the class of linear, unbiased estimators

3.8 Choosing between Pooled OLS and Random effects: Breusch-Pagan Lagrange

Multiplier (LM) Test

This test will check for the random-effects model based on the OLS residual. The LM test

helps to decide between a random effect regression and a simple OLS regression.

𝑯𝟎=Pooled OLS is appropriate than the Random Effects Model.

If P-value > 0.05 then fail to reject the null hypothesis and go for Pooled OLS and If P-
value<

0.05 reject the null hypothesis and go for Random Effects Model.

3.9 Choosing Between Fixed and Random Effects: Hausman Test

The null hypothesis is that the preferred model is random effects vs. the alternative fixed

effects. It tests whether the unique errors (αi) are correlated with the regressors. The null

hypothesis is they are not correlated.

25
Since the random effects estimator is more efficient, it is used if the Hausman test supports it.

If the null hypothesis is rejected, the fixed-effect model will be preferred as the most efficient

model.

3.10 Diagnostic Tests

A regression diagnostic test seeks to evaluate the validity of a model in different ways.

Applying the assumptions of linear regression models, the following tests will be used to

validate the model.

Durbin Watson (DW) test

Durbin Watson (DW) test is used to check for autocorrelation. A correlation matrix shows the

relationship between the dependent and independent variables to ensure no correlation exists

between them.

• 𝐻0=The residuals from an OLS regression are not autocorrelated.

• DW=2 , No autocorrelation

• DW= 0 to <2 ,Positive autocorrelation

• DW= >2 to 4, Negative autocorrelation

Dicky-Fuller & Phillips –Peron Test

Dicky-Fuller & Phillips –Peron test is used to test the stability of the time series of the study

variables. The null hypothesis is that the series has a unit root (no-stationarity). If the unit

root is present, the first difference estimation is used. However, the test is not able to reject

the hypothesis that the data is non – stationary. If p-value < 0.05 then no unit roots present.

26
Breusch- Pagan test.

The Breusch- Pagan test is used to test for homoscedasticity. If heteroscedasticity is detected,

a robust covariance matrix is used to control it.

• 𝐻0=Homoscedasticity is present in the model.

• If P-value <0.05, reject the null hypothesis and conclude heteroscedasticity is present

in the model.

The Breusch –Pagan LM test of independence and the Pasaran CD test is also used to test for

contemporaneous correlation, for instance, cross-sectional dependence.

Jarque -Bera Test

Normality is tested by use of the Jarque – Bera test. This test checks whether the sample

panel data has the skewness and kurtosis that matches a normal distribution. The test is

always non-negative. If it appears to be far from zero, it implies that the sample data does not

have a normal distribution.

Ho: data matches a normal distribution

If P-value <0.05, reject the null hypothesis. The distribution is not normal.

Testing for cross-sectional dependence / contemporaneous correlation: Using Pasaran CD

test

According to Baltagi, cross-sectional dependence is a problem in macro panels with a long time

series. The Pasaran-CD tests whether the residuals are correlated across entities. Cross-sectional

dependence can lead to bias in the test results.

Ho: Residuals across entities are not correlated.

27
CHAPTER FOUR

DATA PRESENTATION, ANALYSIS, AND RESULTS

4.1 Introduction.

This chapter is comprised of data analysis, findings, and interpretation. Results were presented in

tables and diagrams. The analysed data was arranged under themes that reflected the research

objectives.

4.2 Descriptive Analysis of Statistical Data

The descriptive analysis of the dependent and independent variables is presented in Table 4.2 .The

dependent variables are profitability and performance measured by the ROE and ROA respectively,

and the independent variables are: the non-performing loan ratio (NPLR), capital adequacy ratio

(CAR), Non-performing loan ratio (NPLR), the loan loss provision ratio (LLPR), Debt ratio (DR)

and size of the bank (SB). In order to give a brief overview of our data, we present the following

table 4.2 which contains the descriptive statistics of our variables calculated from data obtained in

financial statements of the best performing banks listed on NSE for the period from 2016-2020. The

number of observations for each variable is 120.

28
Table4.2: Descriptive Statistics of the Variables

Variable variables N Mean SD MIN MAX

Dependent ROA 120 0.00734 0.00316 0.00092 0.02654

ROE 120 0.04555 0.01802 0.00470 0.14904

Independent NPLR 120 0.09981 0.03644 0.03970 0.18256

LLPR 120 0.01153 0.01149 0.00117 0.05675

CAR 120 0.31149 0.04059 0.00121 0.24800

DR 120 0.05767 0.07481 0.01990 0.83777

SB 120 13.3200 3.19823 8.05000 19.6100

Source: Research data

The mean value for the dependent variables ROA and ROE is 0.00734 and 0.04555 respectively.

The result shows that the average value of the bank's performance (ROA) is 0.734% indicating that

during the period 2015-2020, on average, the total assets of the banks generated a return of

0.734% .The standard deviation of the ROA is 0.316%, which shows the absence of substantial

variation. The average ROE in the last five years is 4.555%, with standard deviations of 1.802%.

The difference between the minimum value (0.470%) and the maximum (14.904%) and the

standard deviations showed that there was a great variability with the ratio of bad debts. The result,

in general, implies that the accumulation of NPLR that has been claimed as a critical problem of the

banking sector in previous studies (Ogbulu et al., 2016).

The mean scores of independent variables i.e. NPLR, LLPR, CAR, DR, SB are 9.981%, 1.153%,

31.149%, 5.767%, 13.32 respectively. The non-performing loans ratio varies from 3.970% to

18.256% with the average and Standard deviation of 9.981% for all the banks for the period 2020-

2015, and 3.644% respectively, which indicates a high volatility of the bank's ability to manage
29
credit risk. The average capital adequacy ratio is 31.149 %, which is way higher than the regulatory

requirements of 8%. In addition, the banks have an average of 1.153% of the provision for loan

losses with a standard deviation of 1.149%. The maximum and minimum values were respectively

5.675% and 0.117%.

The column of standard deviation denotes how much the variables deviate from the mean, here

NPLR, LLPR, CAR have the lowest standard deviation while SB has the highest standard deviation

of 3.1982, the higher the standard deviation, the higher the variability of that factor.

30
4.3 Analysis of the Evolution of the Dependent and Independent Variables.

Figure 1: Evolution of the Variables

Source: Research Data.

31
Figure 2: Evolution of the Variables

Source: Research Data.

From 2016 to 2019, debt ratio and Capital adequacy ratio exhibit the same trend with a slight
32
increase of Capital Adequacy ratio for the year 2020. As non-performing loans ratio increases, loan

loss provision ratio increases resulting in a decrease in ROE and ROA. In addition, CAR increased

between 2016 and 2020 from all the banks with an exception of I&M and Coop bank. The decrease

can be explained by a decrease in deposits in 2020 or an increase in loans granted to individuals,

which tends to a slight decrease in ROE and ROA.

33
4.3.1 Analysis of the Evolution of Size of the Bank: A Control Variable

DTB
13.2
13.1
13
12.9
12.8
12.7
12.6
12.5
12.4
12.3

2020 2019 2018 2017 2016

KCB
14
13.8
13.6
13.4
13.2
13
12.8

2020 2019 2018 2017 2016

STANCHART
19.65
19.6
19.55
19.5
19.45
19.4
19.35
19.3
19.25
19.2

2020 2019 2018 2017 2016

Figure 3: Evolution of Bank Size

Source: Research Data.

The size of the banks for the six banks is seen to increase from 2016 onwards to 2020 with an

exception of I&M bank which decreased in 2020 and Equity bank which decreased in 2019. This

can be explained by the fact that these banks acquired more customers and this acquisition led to
34
growth of the deposits and the overall size of the bank.

4.4 Box Plots

The box plots presented below were used to show overall patterns of response for a group. They

provide a useful way to visualize the range and other characteristics of responses for a large group.

35
CAR box plot

The box plot is comparatively short and thus suggesting that values for CAR are close to each other

and that we have only one outlier value. Our median value lies between 0.10 and 0.20.

NPLR box plot

The box plot is comparatively tall suggesting that values for NPLR vary from each other, we have

no outliers. Our median value lies between 0.08 and 0.12.

ROA box plot

The box plot is comparatively short implying that values for ROA are close to each other, we have

two outlier values. Our median value lies between 0.00 and 0.001

ROE box plot

The box plot is comparatively short implying that values for ROE are close to each other, we have

three outlier values. Our median value lies between 0.00 and 0.05.

36
SB box plot

The box plot is comparatively short suggesting that values from SB are close to each other, we have

few outlier values. Our median value lies between 10.00 and 14.00.

4.4 Coefficient of Correlation

Table 4.4: Coefficient of Correlation among variables

ROA ROE NPLR CAR DR LLPR SB

ROA 1.0000

ROE 0.9614 1.0000

NPLR -0.1040 -0.1496 1.0000

CAR -0.2702 -0.3050 -0.2854 1.0000

DR 0.0319 0.0240 -0.1811 0.0441 1.000

LLPR -0.2823 -0.2968 0.2882 -0.0252 -0.0448 1.0000

SB -0.0607 -0.0337 0.6139 -0.4124 -0.1772 0.2705 1.0000

Source: Research data

37
The above table demonstrates the coefficient of correlation between the dependent and independent

variables. Independent variables are correlated to ROA and their coefficients of correlation are as

follows 0.9614, -0.1040, -0.2702, 0.0319, -0.2823, -0.0607 the coefficient of correlation between

ROE and the independent variables are as follows -0.1496, -0.3050, 0.0240, -0.2968, -0.0337

respectively. We find that the highest correlation between all variables in absolute value is -0.3050,

which is the correlation between CAR and ROE. This means that a small change in one of the

observations or the number of observations leads to a profound change in the estimated values of

the coefficients.

As can be seen, there is a strong positive correlation between ROA and ROE at 96%. There is also a

weak positive significant correlation between DR and ROA, ROE, and CAR at 3.19%, 2.40%,

4.41% respectively. We also have a strong positive correlation between SB and NPLR, LLPR at

61.39% and 27.05% respectively. There is a negative weak correlation between NPLR, ROA, and

ROE at 10.40% and 14.96%. CAR is also negatively correlated to ROA, ROE, and NPLR at

27.02%, 30.50%, and 28.54 %. LLPR had a negative correlation with all the other variables except

for NPLR.

4.5 Diagnostic Tests

Several diagnostic tests were performed on the data as had earlier been stated in chapter 3. Below

are the listed results.

4.5.1 Heteroscedasticity test: Breusch-Pagan

Breusch Pagan test was used to test for homoscedasticity in the data, the null hypothesis for the test

was that there is a presence of homoscedasticity. Below are the results

Table 4.5.1.1: Heteroscedasticity Test

BP df P-value

ROA 46.624 5 0.000836

ROE 35.495 5 0.00297


38
Source: Research data

Our p-values for both ROA and ROE were less than 0.05, therefore we reject the null hypothesis

and conclude that there is heteroscedasticity in the data.

As a solution to the presence of heteroscedasticity, Robust Standard errors were utilized to solve its

presence in the data. The function VcovHc () was used to estimate a heteroscedastic consistent (HC)

variance-covariance matrix for the parameters, below are the results for the regression model that

utilized the heteroscedastic-consistent estimation of the covariance matrix as the solution to the

heteroscedasticity present in the data.

Table:4.5.1.3 Heteroscedastic Consistent Variance-covariance matrix.

intercept lnNPLR LLPR lnCAR lnDR SB

inercept 0.00001603

lnNPLR 0.00000167 0.00000035

LLPR -0.00000216 0.00000046 0.00033996

lnCAR 0.00000180 0.00000011 -0.0000070 0.0000006

lnDR 0.00000078 0.00000009 0.00000206 -0.00000002 0.00000026

SB -0.00000052 -0.00000003 -0.00000037 -0.00000004 0.00000001 0.00000003

Source: Research data

ROE results:

Table: 4.5.1.2 ROE Heteroscedastic Test Results

Estimate Std. Error t value Pr(>|t|)

intercept 0.01269817 0.02269521 0.5595 0.5769122

lnNPLR -0.0118689 0.00348484 -3.4059 0.0009116

39
LLPR -0.3928214 0.11949730 -3.2873 0.0013452

lnCAR -0.0072616 0.00063261 -11.4789 0.00000

lnDR 0.00270278 0.00535043 0.5052 0.6144275

SB 0.00031120 0.00071902 0.4328 0.6659699

Source: Research data

Table: 4.5.1.3 ROA Heteroscedastic Test Results

Estimate Std. Error t value Pr(>|t|)

intercept 0.145134 0.179910 0.8067 0.421518

lnNPLR -0.083099 0.150217 -2.0351 0.044161

LLPR -0.882399 -0.327860 - -2.6916 0.008182

lnCAR -0.066496 0.018410 -13.2868 0.00000

lnDR 0.049869 0.110527 0.4512 0.652703

SB -0.014922 0.024431 -0.1660 0.868428

Source: Research data

From the above results a conclusion was made that after the use of the Heteroscedastic-Consistent

estimation of the covariance matrix, it is evident that the values of our standard errors do not differ

that much in comparison to the random-effects model, leading us to the conclusion that the presence

of heteroscedasticity in our data is not significant enough to change the efficiency of our estimates.

Also, the sample size of our data could be sufficiently large enough to the extent that the variance of

our estimators is still small enough to get precise estimates.

4.5.2 Serial- Correlation test

The Durbin-Watson test for serial correlation was utilized. The null hypothesis for the test was no

serial correlation in idiosyncratic errors. Below are the results

Table: 4.5.2 Serial Correlation Test Results


40
DW P-value

ROA 2.0279 0.5666

ROE 2.0806 0.4511

Source: Research data

The p-values obtained are higher than 0.05 hence we fail to reject the null hypothesis and conclude

that there is no serial correlation in the data.

4.5.3 Stationarity test

To enable efficient predictions using our data, it had to be stationary. Stationarity tests were done

for each variable and three independent variables were found to be non-stationary in our data.

Levin-Lin-Chu unit root tests were done to check for stationarity the null hypothesis for the test was

the presence of a unit root, below are the results.

Results for the three variables were found to be non-stationary.

Table: 4.5.3.1 Levin-Lin-Chu Unit Root Test Results

Independent Variable Z P-value

NPLR -0.23553 0.4069

CAR -1.1274 0.1298

DR 0.01769 0.5071

Source: Research data

All three variables had a p-value greater than 0.05. As a result, we fail to reject the null hypothesis

and conclude that the variables are non -stationary.

To address the issue of non-stationarity, log-transformed variables were used to control non -

stationarity. Dickey-Fuller test for unit root was used to check for non-stationarity. The null

hypothesis was that unit root is present. Below are the results;

Table: 4.5.3.2 Dickey-Fuller test


41
Dickey-fuller Lag p-value

NPLR -5.2767 4 0.01

CAR -4.8458 4 0.01

DR -9.3591 4 0.01

Source: Research data

From the above results, the p-value for all three values is less than 0.05, and thus we reject the null

hypothesis and conclude that the three variables are stationary.

4.5.4 Poolability Test

Test for poolability was done to check whether the pooled OLS regression would provide efficient

estimates. The null hypothesis for the test states that the Pooled OLS is an efficient estimator.

Below are the results.

Table: 4.5.4 Poolabilty Test Results

F P-value

ROA 1.5261 0.07915

ROE 1.2053 0.2597

Source: Research data

Both ROA and ROE had a p-value greater than 0.05. Therefore, fail to reject the null hypothesis and

conclude that a pooled OLS is an efficient estimator.

To choose the most efficient model among the three, Pooled OLS, Fixed Effects, and Random

Effects two more tests were done. The first test is the Breusch Pagan Lagrange Multiplier test to

choose between the pooled OLS and Fixed Effects Models. The second test was the Hausman test

that was used to choose between the Random Effects model and the fixed effects model.
42
Below, we post the results for the two tests.

4.6.1 Breusch Pagan Lagrange Multiplier test.

The test is used to choose between the Pooled OLS and Random Effects Models.

Table 4.6.1 Breusch Pagan Lagrange Multiplier test

Dependent Variable F P-value

ROA 7.5767 0.009553

ROE 5.8641 0.052232

Source: Research data

The null hypothesis for this test was that the pooled OLS is the efficient model. From the results,

the p-values for the two variables were less than 0.05. Therefore, we reject the null hypothesis and

conclude that the Random Effects model is the efficient model compared to the Pooled OLS.

4.6.2 Hausman test

The test is used to choose between the Random Effects model and the fixed effects model.

Table 4.6.2 Hausman Test Results

Chisq P-value

ROA 4.4847 0.4819

ROE 5.0949 0.4044

Source: Research data

The null hypothesis for the test was that the Random Effects model was the efficient model. From

our results, the p-value for the two variables was greater than 0.05. Therefore, we fail to reject the

null hypothesis and conclude that the Random Effects model is the efficient model between the two.

From, the above two tests it is evident that the Random Effects model is the most efficient model to

be used to estimate and fit the model.

The Random Effects model was fitted and below we post the results obtained from fitting the

model:
43
4.7 Discussion of the Results

4.7.1 Regression model 1: ROA

Table: 4.7.1.1 Regression results: ROA

Estimate Std. Error z-value Pr(>|z|)

intercept 0.145134 1.188437 0.7532 0.451350

ln (NPLR) -0.083099 -0.144027 -2.1226 0.033790 *

LLPR -0.882400 0.297156 -2.9695 0.002983 **

ln (CAR) -0.066495 0.051795 -4.7228 0.005768***

ln (DR) 0.049869 0.114125 0.4370 0.662130

SB -0.014922 0.0311585 -0.1302 0.896420

Source: Research data

The estimated equation under the Random Effects model is:

ROA= 0.14513 intercept -0.08309 ln (NPLR) -0.88240 LLPR - 0.06640 ln (CAR) + 0.04987 ln

(DR) -0.01492 SB.

The results indicate that the Debt Ratio (DR) had the highest positive relationship with Return on

Assets. This implies that a change in one unit of DR will result in the improvement of 4.98% on

Return on Assets. For the rest of the variables, a change in one unit leads to a negative change in the

Return on Asset. The p-value indicates the significance of the independent variables in explaining

the dependent variable. If the significance number found is less than the critical value also known as

the probability value (P) which is statistically set at 0.05. Then the conclusion would be that the

model is significant in explaining the relationship, else the model would be regarded as non-

significant (Bagozzi, 1988). From our above results there were three significant variables; Non-

performing Loan Ratios, Loan Loss Provision, Credit Adequacy Ratio with values 0.033790,

0.002983, 0.005768 .Two variables were non-significant: Debt Ratio and Size of the bank.
44
Table: 4.7.1.2 Model Fitness

Total Sum of Squares 0.0010181

Residual sum of squares 0.00071819

R-Squared 0.2946

Adj. R-Squared 0.26366

Source: Research data

The results presented in the above table show the fitness of the model as used in the regression

model to explain the study phenomena. Non-performing Loan Ratios, Loan Loss Provision, Credit

Adequacy Ratio, Debt Ratio, and Size of the bank provide a moderately good fit in predicting

changes in the financial performance of commercial banks. The model had an R-squared of 29.46%

and an adjusted R-squared of 26.366%. This means that the Non-performing Loan Ratios, Loan

Loss Provision, Credit Adequacy Ratio, Debt Ratio, and Size of the bank explain 29.46 of the

variation in the dependent variable, which is the profitability of commercial banks explained by the

ROA.

Discussion on ROA:

Bank Size

Based on Pearson’s correlation, table 14.10 (first table under Findings on ROA) shows that there is

a negative relationship between the size of the bank and the ROA. Equally, the variable is less

significant to explaining to the model since it has a p-value of 0.896, which exceeds 0.05. When all

other factors are held constant, a 1% increase in SB will lead to a decrease of 0.0149225 in the

performance of commercial institutions.

Non –Performing Loans Ratio

From table 14.10, there is a negative relationship between the NPLR and the ROA. However, the

variable is significant to explaining to the model since it has a p-value of 0.0338, which is below

0.05. When all other factors are held constant, a 1% increase in NPLR will lead to a decrease of
45
0.083116 in the performance of commercial institutions.

Capital Adequacy Ratio

Concerning table 14.10, CAR has a negative relationship with the ROA. However, it is worth noting

that the variable is significant to explain the model. This is justified by having a p-value of 0.0058,

which is below 0.05. When all other factors are held constant, a 1% increase in NPLR will lead to a

decrease of 0.066496 in the performance of commercial institutions.

Debt Ratio

Table 14.10, implies the existence of a positive relationship between ROA and the Debt Ratio

(CAR), even though the variable is less significant in explaining the mode. This is so because the

variable has a p-value of 0.6621 that is beyond the statistically set 0.05. A 1% increase in DR will

result in a decrease in the performance of commercial institutions by 0.0149225.

Loan Loss Provision.

The tabulated results on Pearson’s Correlation coefficients as presented in table 14.10 show that

there is a negative relationship between LLPR and the performance of commercial institutions,

concerning the ROA. Nevertheless, the variable is significant in explaining to the model as can be

seen by its p-value of 0.0029, which falls below the statistically set 0.05. A 1% increase in LLPR

will result in a decrease in the performance of commercial institutions by 0.882316.

4.7.2 Regression Model 2: ROE


46
Table: 4.7.2.1 Regression results: ROE

coefficients Estimate Std. Error z-value Pr(>|z|)

(Intercept) 0.01735853 0.02128392 0.8156 0.414746

NPR -0.01153641 0.00466401 -2.4735 0.013380 *

LLPR -0.36469183 0.13220492 -2.7585 0.005806 **

ln (CAR) -0.00609536 0.00154833 -3.9367 8.26e-05 ***

ln (DR) 0.00250625 0.00365696 0.6853 0.493132

SB 0.00012753 0.00071961 0.1772 0.859335

Source: Research data

Thus, the optimal model for the study is;

ROE = 0.01735853 + (-0.01153641) ln (NPLR) + (-0.36469183) LLPR + (-0.00609536) ln (CAR)

+ 0.00250625 ln (DR) + 0.00012753 SB

The results indicate that the Debt Ratio has the highest impact on the Return on Equity of

commercial banks with a change of one unit leading to an improvement of 0.25% of the Debt ratio.

A change in a unit of size of the bank will lead to an improvement of 0.0127% of the Return on

Equity, while a one-unit change in loan loss provision ratio will lead to a negative change of 36% of

the Return on Equity. A unit change of capital adequacy ratio will lead to a negative change of

0.61% of the Return on Equity of a bank. Lastly, a unit change of the non-performing loans ratio

will lead to a negative change of 1.15% of the return on Equity.

In statistics, the p-value indicates the level of the relation of the independent variable to the

dependent variable. If the significance number found is less than the critical value also known as the

probability value (p) which is statistically set at 0.05. Then the conclusion would be that the model

is significant in explaining the relationship else the model would be regarded as non-significant

(Bagozzi, 1988). In our data, the Non-Performing Loans Ratio, the Loan Loss Provision Ratio, and
47
the Capital Adequacy Ratio are significant in explaining the dependent variable ROE since their p-

values are less than 0.05.

Table: 4.7.2.2 Model Fitness

Total Sum of Squares 0.035591

Residual sum of squares 0.025856

R-Squared 0.24168

Adj R-Squared 0.24168

Source: Research data

The results presented in the above table present the fitness of the model used in the regression

model to explain the Return on Equity. The size of the bank, loan loss provision ratio, non-

performing loans ratio and capital adequacy ratio provide a moderately good fit in predicting

changes in the Return on Equity of banks which shows the profitability of banks. The model has an

R-Squared of 0.24168 and an Adjusted R-Squared of 0.24168 which means that the Size of the

Bank, Loan Loss Provision Ratio, Non-Performing Loans, and Capital Adequacy Ratio explain

24.17% of the variations in the dependent variable which is the Return on equity of commercial

banks. These results further imply that the model for the relationship of the variables provided a

relatively good fit.

The results of the panel data analysis on ROE are presented in Table 4.7.2.1 and discussed below.

Table 4.7.2.1 summarizes the outcome of the random effect model using ROE as the dependent

variable. The Hausman test for ROE calculated the p-value to be 0.4819 which is greater than the

alpha of 0.05, implying that the random effect model is a better model specification than the fixed

effect model. Under the random effect model, the variations across entities are assumed to be

random and uncorrelated with the independent variables.

The Random Effects model on ROE shows that the independent variables NPLR, LLPR, and CAR
48
are the only variables that are statistically significant in explaining ROE. On the other hand, the

other remaining independent variables; SB and DR are not statistically significant in explaining

ROE.

Discussion on ROE.

Bank size

Bank size is found to be significantly affecting the conventional banks’ credit risk in Kenya and a

negative relationship existed between bank size and credit risk. This inverse relationship implies

that the larger the bank size the lower the credit risk is. This result is similar to the findings by Salas

and Saurina (2002) and Hu et al., (2004) who also found that bank size and credit risk have a

negative relationship. While Bardhan and Mukherjee (2016) establish that larger banks are more

exposed to default risk than smaller banks. Besides that, larger banks will have more capital to

buffer the risk related to credit as compared to smaller size banks.

The results presented in Table 4.7.2.1 indicates that SB and ROE have a negative relationship and

that SB is not statistically significant at a 5 percent level of significance. Keeping all the other

factors constant, a 1 percent increase in size of the bank leads to an increase in ROE by 0.012753

percent. From the above results, it can be concluded that an increase in size of the bank leads to an

increase in performance.

Non –Performing Loans Ratio

Non-Performing Loan Ratio (NPLR) is identified as an important factor in managing credit risk in

the banks. It is hypothesized to have a negative relationship with bank profitability. The result in

Table 14.2 indicates that NPLR is negative and statistically significant at a 5 percent level of

significance which confirms the expectations. Further, the result shows that NPLR indeed

influences the ROE of Kenyan commercial banks. Keeping all the other factors constant, a 1 percent

increase in NPLR leads to a decrease in ROA by 0.1107 percent.


49
Capital Adequacy Ratio

Capital Adequacy Ratio (CAR) in this study is used as an independent variable and is hypothesized

to have a positive relationship with bank profitability. On the contrary, the result in Table 4.7.2.1

shows that CAR is negative and statistically significant in explaining the ROE at a 5 percent level of

significance as expected. It indicates that CAR is also an important variable affecting the ROE of

the 6 commercial banks under study. Keeping all the other factors constant, a 1 percent increase in

CAR leads to a decrease in ROE by -0.00609536.

Debt Ratio

The result in Table 4.7.2.1 indicates that DR is positively correlated with ROE but not statistically

significant in explaining the dependent variable at a 5 percent level of significance. A 1 percent

increase in DR leads to a 0.00250625 percent increase in ROE.

Loan Loss Provision.

From our results obtained LLPR is negatively significant at a 5% level of significance, this shows

that LLPR affects the profitability of a bank; for one unit increase in LLPR, there is a decrease in

the profitability of the banks. If seen from a logical point of view an increase in loan loss provision

leads to a decrease in banks profitability, since funds needed for investments are utilized to establish

a cushion in case of losses suffered from loans.

50
Analysis of Variance: ROE

df F-value P-value

lnNPLR 1 3.755 0.0551

LLPR 1 10.730 0.0014

lnCAR 1 22.493 0.0152

lnDR 1 2.123 0.1478

SB 1 0.005 0.9415

Residuals 114

The null hypothesis states that the mean values of our five variables are equal. In our case for the

variables lnNPLR, lnDR and SB have P-values greater than our significance value of 0.05.

Therefore we fail to reject the null hypothesis and conclude that the above-mentioned variables

have equal means.

For, the two remaining variables LLPR and lnCAR they have a p-value of less than our significance

value therefore we reject the null hypothesis and conclude that the two variables have different

means.

Analysis of Variance: ROA

df F-value P-value

lnNPLR 1 1.583 0.21087

LLPR 1 10.807 0.00134

lnCAR 1 23.130 0.01157

lnDR 1 4.007 0.04769

SB 1 0.501 0.48042

Residuals 114

51
From the above data two variables, lnNPLR and SB had a p-value greater than our significant value

of 0.05. We fail to reject our null hypothesis (that the mean value of our variables is equal) and

conclude that for the two mentioned variables their mean values are equal. For the rest of the

variables LLPR, lnCAR and ln DR had a p-value less than our significant value of 0.05. We reject

the null hypothesis and conclude that the three variables have different mean values.

52
CHAPTER FIVE

CONCLUSIONS AND RECOMMENDATIONS

5.1 Conclusions

At the beginning of our research, it is explained that the goal of this paper is to analyse the

relationship between credit risk management indicators, financial performance, and profitability of

commercial banks in Kenya. This was done by collecting data from 6 Kenyan commercial banks

from 2016 to 2020. To determine the relationship between credit risk management, financial

performance, and profitability, we chose ROE and ROA as the proxies for bank profitability and

financial performance, and CAR, NPLR, LLPR, DR, and DR as proxies for credit risk management.

After the data collection, R (a statistic program) was used to test for the research questions. For this

study, a panel data analysis was applied for the 6 selected commercial banks of Kenya for the

period 2016 to 2020. The random-effect model was the consistent model and used for the analysis.

According to the empirical findings of this study, the conclusion was that there exists a relationship

between credit risk management and profitability of Kenyan Commercial banks from the period of

2016 to 2020.

To begin with, the findings reveal that NPLR is negatively correlated with both ROE and ROA.

This is consistent with most of the previous relating researches. This relationship signifies that

having a high number of non-performing loans in a portfolio reduces the financial performance and

profitability of commercial banks since there is less capital for the commercial banks to invest and

operate. Secondly, CAR is significant and negatively correlated with both ROA and ROE. This

suggests that the bank has a lot of capital to carry out its operational activities and can bear the risk

if the bank is liquidated. Therefore we cannot sufficiently conclude that an increase in CAR reduces

the financial performance and profitability of the commercial banks.

The size of the bank also has a negative correlation with both ROE and ROA. It is not significant in

explaining the financial performance and profitability of commercial banks. This means that the
53
small and large banks can enjoy profitability influenced by factors other than their size. In addition,

the Debt Ratio is not significant and has a positive correlation with both ROE and ROA. This means

that banks can increase their leverage for better performance. LLPR was significant and had a

negative correlation between ROE and ROA. This implies that the selected banks have set aside a

huge amount of capital for cushion purposes instead of using such an amount to invest and operate

their businesses. This would enable them to mitigate the credit risk as it arises.

In summary, the conclusion is that there is a positive relationship between credit risk management

and the profitability of commercial banks in Kenya. The better and more effective the credit risk

management, the less the exposure to credit risk and the higher the profitability to the commercial

banks.

5.2 Recommendations

The following suggestions are derived from the research findings:-

 For commercial Banks to design an effective credit risk management system, they need to

establish a suitable credit risk environment; operate under a sound credit granting process,

maintain an appropriate credit administration that involves monitoring, processing as well as

enough controls over credit risk.

 Even though the Debt ratio is insignificant in explaining the financial performance and

profitability of commercial banks in our study, banks should pay much attention to its trend

since leveraging a bank means improving its performance in general.

 Banks should also ensure they have enough cushion to absorb a reasonable amount of losses

before they become insolvent keeping an eye on their investment strategies to ensure the

operations generate profitable cash flows to also cater for the losses too in case they occur.

 Banks must adhere to prudential banking practices and even diversify more in lending to

54
avoid repetitive losses in particular types of loans issued to their customers.

5.3 Areas for Further Research

A suggestion for further research could be performing research on the relationship between

financial risk management and financial performance of Kenyan commercial banks focusing on

other risk management such as liquidity risk, market risk, or operational risk.

Another area of research could be the inclusion of microfinance banks, Sacco’s and other finance

companies, and cooperatives that are successfully operating in the Kenyan market.

55
REFERENCES

Abbas, F., Iqbal, S., & Aziz, B. (2019). The impact of bank capital, bank liquidity & credit risk on

profitability in post-crisis period: A comparative study of US & Asia. Cogent Economics &

Finance, 7(1), 1-18.

Afriyie, H. O., &Akotey, J. O. (2012). Credit risk management and profitability of selected rural

banks in Ghana. Ghana: Catholic University College of Ghana.

AGYEI, S. K. (2010). Capital Structure and Bank Performance in Ghana. The University of Ghana.

Unpublished Thesis.

Aktan, B. and C. Bulut. (2008). Financial performance impacts of corporate entrepreneurship in

emerging markets: A case study of Turkey. European Journal of Economics, Finance and

Administrative Science, 12: 69-79.

Altunbas, Y. (2005). Mergers and acquisitions and bank performance in Europe – The role of

strategic similarities. European central bank, Working Paper Size Series, No. 398.

Anthony, M.C. (1997). Commercial bank risk management; an analysis of the process. The

Wharton school. The University of Pennsylvania. Financial Institutions Centres.

Appa, R., (1996). The monetary and financial system. 3rd Edn: London Bonkers Books Ltd.

Athanasoglou, P., S.N. Brissimis and M.D. Delis. (2005). Bank-specific, industry-specific and

macroeconomic determinants of bank profitability. MPRA Paper No. 153.

Baltagi, Badi H. (2005). Econometric Analysis of Panel Data. England: John Wiley & Sons.

Basel Committee on Banking Supervision. (2006). Studies on credit risk concentration, An

overview of the issues, and a synopsis of the results from the research task force project.

Available from www.bis.org.

56
Basel Committee on Banking Supervision. (1982). Management of banks’ international lending,

country risk analysis, and country exposure measurement and control. Available from www.bis.org.

Bessis, J. (2002). Risk management in banking. John Wiley & Sons.

Bobakovia, I.V. (2003). Raising the profitability of commercial banks, BIOTEC, Retrieved on

April 2005, 11. Available from http/www/nbs.SK/BIATEC/.

Boffey, R. and Robson, G. (1995). Bank Credit Risk Management, Journal of Managerial

Finance, vol. 21, no.1, pp 66-78). https://doi.org/10.1108/eb018497.

Bourke, P. (1989). Concentration and other determinants of bank profitability in Europe, North

America, and Australia. Journal of Banking and Finance, 13(1): 65–79.

Campbell, A . (2007). Bank insolvency and the problem of non-performing loans. Journal of

Banking Regulation, 9(1): 25– 45.

Chen, T. and Yeh, T. (1997). A Study of Efficiency Evaluation in Taiwan’s Banks.

https://doi.org/10.1108/09564239810238820.

Collins, N. H., and Preston, L. E. 1969. Price-cost margins and industry structure.’’Review, 51,

August, 271–287.

Colquitt, J. (2007). Credit Risk Management. How to Avoid Lending Disasters and Maximise

Earnings, 3rd Edition, McGraw Hill.

Denscombe, M. (2007). The Good Research Guide, New York: Prentice-Hall.

DeYoung, R. and G. Whalen. (1994). Banking industry consolidation: Efficiency issues, Working

Paper No. 110. A Conference of the Jerome Levy Economics Institute, Office of the

Comptroller of the Currency, Washington, DC.

Felix, A.T. and T.N. Claudine. (2008). Bank performance and credit risk management, unpublished

master dissertation in finance, University of Skovde. Available from

http//www.essays.se/essay/55d5c0bd4/.
57
Gieseche, K. (2004). Credit risk modeling and valuation: An introduction, credit risk: Models and

Management. Cornell University, London.

58
Haselmann & Wachtel. (2006). Bank Risk and Bank Management in Transition: A progress report

on the EBRD banking environment and performance survey.

Heffernan, S. (1996). Modern banking in theory and practice. Chichester: John Wiley and Sons.

Hosna, A., Bakaeva, M., &Juanjuan, S. (2009). Credit Risk Management and Profitability in

Commercial Banks.Master Degree Project No. 2009:36, Gothenburg: School of Business,

Economics, and Law, University of Gothenburg. Retrieved from

https://gupea.ub.gu.se/handle/2077/20857

Houser, J. (2011). Book Alone: Nursing Research, Sudbury: Jones & Bartlett Publishers. Ho, C. and

Zhu, D. (2004). Performance Measurement of Taiwan’s Commercial Banks.

Kargi, H.S. (2011). Credit Risk and the Performance of Nigerian Banks.

Kim, D. and A.M. Santomero. (1988). Risk in banking and capital regulation. Journal of Finance,

43(5): 1219-1233.

Kithinji, A. M. (2010). Credit risk management and profitability of commercial banks in

Kenya. School of Business, University of Nairobi, Nairobi.

Koehn, M. and A.M. Santomero. (1980). Regulation of bank capital and portfolio risk.

Journal of Finance, 35(5): 1235-1244.

Kolapo, T.F., R.K. Ayeni and M.O. Oke. (2012). Credit risk and commercial banks’ performance in

Nigeria: A panel model approach. Australian Journal of Business and Management

Research, 2(2): 31-38.

Lei, H. (2005). The Determinants Impacting Bank Performance: A research dissertation on Two

Banks – HSBC Holding Plc and Royal Bank of Scotland. Masters Degree Project,

University of Bradford.

Mathuva, D. (2009). Capital Adequacy, Cost Income Ratio and Performance of Commercial Banks:

The Kenyan Scenario, the International Journal of Applied Economics and Finance, vol 3,
59
no. 2, pp 35-47. https://doi.org/10.3923/ijaef.2009.35.47.

60
Matthews, T. D. & Kostelis, K. T. (2011). Designing and Conducting Research in Health and

Human Performance, New York: John Wiley & Sons.

Miller, S.M. and A.G. Noulas. (1997). Portfolio mix and large-bank profitability in the USA.

Applied Economics, 29(4): 505- 512.

Mugenda, O.M. &Mugenda, A.G. (2003). Research Method Quantitative and Qualitative

Approaches: Nairobi Kenya; Act Press.

Oberholzer, M. and Westhuizen, G. (2004). An Empirical Study on Measuring Efficiency and

Profitability of Bank Regions, Available at

http://www.meditari.org.za/docs/2004VOL1/165_178.pdf. Accessed 22 June 2016.

Ogboi, C., & Unuafe, K. (2013). Impact of CR management & capital adequacy on the financial

performance of commercial banks in Nigeria. Journal of Emerging Issues in Economics,

Finance & Banking, 2(3), 703- 717

Owojori, A.A., I.R. Akintoye, and F.A. Adidu. (2011). The challenge of risk management in

Nigerian banks in the post-consolidation era. Journal of Accounting and Taxation, 3(2): 23-

31.

Philip, D. (1994). The panacea to distress in banking, a paper presented at the UBA management

retreat.

Robert, D. and W. Gary. (1994). Banking industry consolidation: Efficiency issues, working paper

No 110 presented at the financial system in the decade ahead: A Conference of the Jerome

Levy Economics Institute April: 14 – 16.

Saunders, M., Lewis, P. & Thornhill, A. (2009). Research Methods for Business Students. (5th ed.)

Harlow: FT/Prentice Hall.

Sanusi, J. (2002). The central bank and the macroeconomic environment in Nigeria.

Quarterly Review, 3(3): 25-34.

61
Shampoo, A. E, & Resnik, D. B. Responsible conduct of research .2003.

62
Sylvie de C., & Gautier,B. (2010). Gestion de la banque du diagnostic à la stratégie(6th éd., p.

294).

Yuga, R. B. (2016). Effect of Credit Risk on the Performance of Nepalese Commercial

Banks. NRB Economic Review,42-64.

Umoh, P.N. (2002). Bank loans, recovery: The roles of the regulatory/supervisory authorities.

Judiciary Law Enforcement Agencies and the Press. NDIC Quarterly, 4(3): 8.

Wet, J. and Toit, E. (2006). Return on Equity: A Popular but Flawed Measure of Corporate

Financial Performance. South African Journal of Business Management, vol 38, no.1, pp 59-

69.

Zikmund, W.G., Babin, B.J., Carr, J.C., & Griffin, M (2013) Business Research Methods. (engage

learning).

63
APPENDICES

Workplan

ACTIVITY APRIL MAY JUNE JULY AUGUST

2021 2021 2021 2021 2021

Conceptualization of

the proposal

Development of

proposal

Data collection

Data entry

Data analysis

Data interpretation

Project presentation

64
Budget Estimation

CATEGORY REMARKS UNITS UNIT COSTS TOTAL COST

(KSHS) (KSHS)

Proposal Printing drafts 300 5 1500

development

Stationary 4 20 80

Proposal copies 3 copies 200 600

Data collection Data abstraction 50 10 500

forms

Transport 10 100 1000

stationary 10 20 200

Total 3880

R-Codes
65
66

You might also like