The impact of credit risk on the profitability of commercial banks in vietnam

MINISTRY OF EDUCATION AND BANKING STATE BANK OF VIETNAM HO CHI MINH UNIVERSITY OF BANKING GRADUATION THESIS THE IMPACT OF CREDIT RISK ON THE PROFITABILITY OF COMMERCIAL BANKS IN VIETNA

INTRODUCTION TO THE RESEARCH TOPIC

Reasons for choosing the topic

Through the financial services they provide, banks are closely linked to economic growth The intermediary role of banks is being considered as an engine of economic growth The stability of the banking sector is thus considered to be a prerequisite for macroeconomic stability and growth Profitability and capital adequacy are crucial for the stability of the banking sector

Due to their dynamic structure and the complex nature of the economy in which they are active, banks face a number of risks According to Koch and MacDonald (2000) , the risks faced by banks can be classified into 6 categories which are Credit risk, liquidity risk, market risk, operating risk, nominal risk and legal risks The profitability, market value, liabilities and equity of financial institutions may be negatively affected by each of these risks Loans granted by commercial banks are the main source of income for the banking sector Consequently, credit risk is one of the biggest risks that banks have to face The Basel Committee on Banking Supervision (2011) defines credit risk as the probability of a partial or total loss in respect of an outstanding loan due to late payments The minimal cost of debt and equity is increased by the increase in credit risks Correspondingly, the cost of bank’s funding increases The tendency of banks to experience financial crises is increasing as their exposure to credit risks increases

After the Global Financial Crisis occurs in 2008, despite the relatively limited exposure of the Vietnamese financial and banking field due to its early stage of integration into the global financial system, the direct effect of the financial crisis led to a reduction in profits for many banks, and even resulted in losses for some smaller- scale banks Consequently, the number of non-performing loans rose, which had an impact on the profitability of the Vietnamese commercial banking sector in the following years

Credit risk in the banking industry is a significantly interest area for researchers and policymakers In Vietnam, following a period of quick credit growth with numerous revealed risks, since 2012, the State Bank of Vietnam (SBV) has begun to tightly control the issue of credit growth in the banking industry and sees it as an important tool in monetary policy management Since then, the SBV has used the financial standing and operational actions of individual commercial banks as the foundation for setting yearly goals for credit expansion Commercial banks have recently raised the percentage of non-interest income and broadened the composition of their revenue streams Nonetheless, ending remains to be the primary commercial activity for banks due to the intermediary nature of financial transactions in banking operations, with credit risk playing a crucial role in the health and functioning of banks Effectively controlling credit risk at the national level will greatly support economic growth, while at the bank level, it will enable them to achieve substantial profits from their core business activities

The objective of this study is to empirically investigate the impact of credit risk (CR) on the profitability of VCBs during the period from 2013 onwards, a time characterized by pronounced divergence in profitability among large-scale and smaller-scale banks This period witnessed modest financial performance of commercial banks, amidst challenging economic conditions, posing significant industry-wide challenges with various fluctuations (in terms of financial structure, business efficiency, and profitability)

The study seeks to examine how credit risk, as represented by various indicators such as non-performing loans and credit quality, influences the profitability of these banks Understanding this relationship is paramount as it provides insights into the dynamics affecting bank profitability amid adverse economic conditions and fluctuating market conditions By unraveling the nexus between credit risk and profitability, policymakers and industry stakeholders can devise informed strategies

3 to enhance the financial performance and resilience of VCBs in the face of evolving market challenges

Vietnam's economy is experiencing rapid development and deep integration into the global economy With the current trend of economic integration in Vietnam, the banking system plays a crucial role in channeling capital from surplus areas to areas with capital shortages Additionally, banks provide essential financial services to individuals, organizations, and the government However, Vietnamese Commercial Banks (VCB) are increasingly facing challenges stemming from credit risk associated with lending to customers Therefore, the effectiveness of business operations becomes a particularly important criterion, with profitability being the most crucial goal, determining the survival and development of banks

In recent years, non-performing loans (NPLs) have not only been a "disease" of the Vietnamese banking system but have also become a concerning issue for the global banking and financial system, especially after the global financial crisis of

2008 originating from the United States, spreading to neighboring countries and worldwide The term "non-performing loans" has become a topic addressed in many studies in developed and emerging countries, including Vietnam

Non-performing loans are understood as debts that exceed the repayment deadline by a certain number of days, and borrowers are unable to fulfill their repayment obligations to the lending bank This is a credit risk commonly encountered by Commercial Banks NPLs violate the fundamental characteristics of credit, namely timeliness and full repayment Furthermore, they erode the trust of creditors in borrowers receiving credit The emergence of NPLs leading to credit risk is unavoidable, especially in the banking sector This is a challenge that all VCBs must face; if the NPL ratio is too high, bank operations will be paralyzed because banks lack capital to repay depositors when due At a severe level, this can lead to

4 bank bankruptcies Therefore, credit risk management, prevention, limitation, and handling of NPLs are crucial tasks for Commercial Banks

In the majority of previous empirical studies on the factors through which credit risk affects profitability, macroeconomic or bank-specific factors were considered separately There have been a few studies that combined both groups of factors in an independent study Moreover, most empirical literature analyzing credit risk factors has focused on developed countries and some typical developing countries, while only a few studies have concentrated on a rapidly growing emerging market with a relatively small scale economy like Vietnam

The relatively limited research and references on the impact of credit risk on the profitability of Commercial Banks in Vietnam have been one of the driving factors behind conducting this study The selection of Commercial Banks as the research subject is based on the following variables Firstly, through the Return on Assets (ROA) ratio, which is commonly used to measure the profitability of an investment relative to the average total assets of the bank ROA is a crucial ratio for accurately assessing the actual profits earned by Commercial Banks, as well as demonstrating the potential development of the banking sector in Vietnam from 2013 to 2023 Secondly, based on the statistics of the increase in Non – Performing Loans (NPLs) of Vietnamese Commercial Banks from 2013-2023 The increase in NPLs also leads to higher credit risk, which is a factor contributing to the risk of bankruptcy Thirdly, the concern over the fluctuation of liquidity and the ratio of capital adequacy to the loan-to-deposit ratio in Commercial Banks based on the statistics of Vietnamese Commercial Banks (VCBs) from 2013-2023 Therefore, analyzing the impact of credit risk on the profitability of banks is extremely necessary

Research Objective

The research aims to measure the impacts of credit risk on the profitability of Vietnam Commercial Banks (VCBs) and propose recommendation to mitigate credit risk in order to enhance the profitability of these Commercial Banks in Vietnam

• the impacts of credit risk on the profitability of Vietnamese commercial banks and propose recommendations to mitigate credit risk to enhance the profitability of these banks

• Identify and measure the credit risks and profitability of commercial banks; analyze the theoretical and empirical basis of the impact of credit risk on the profit and profitability ratios of these banks

• Assess the level of influence of credit risk on the profitability of commercial banks in Vietnam; provide recommendations for solutions to mitigate credit risk to enhance earning capacity and deliver tangible profits for these banks.

Research Questions

The research has the following research questions:

Question 1: Does credit risk have an impact on the profitability of Vietnamese

Question 2: What is the extent of the influence of credit risk on the profitability of

Question 3: What are the recommendations to enhance the profitability of

Research Subject and Range

The research subjects of the topic are Vietnamese Commercial Banks from

2013 to 2023, based on financial statements and annual reports

The research investigates the impact of credit risk on the profitability of twenty

(20) Vietnamese commercial banks from 2013 to 2023 and proposes mitigation strategies The period from 2013 to 2023 was chosen to ensure a sufficiently long and recent timeframe for a comprehensive and accurate study of the impact of credit risk on the profitability of Vietnamese commercial banks This period includes significant economic fluctuations, such as the economic recovery following the global financial crisis, with stable GDP growth of around 5% to 7% annually The Vietnamese banking sector also experienced robust development, exemplified by increased capitalization and improved asset quality Simultaneously, numerous changes in financial policies and banking regulations were implemented, such as the adoption of Basel II in 2020 and regulations on capital adequacy ratios, which have altered the operational environment and risk management practices of commercial banks These factors create a rich and diverse context for studying the influence of credit risk on bank profitability over the past decade.

Research Methodology

This study investigates the impact of credit risk on the profitability of commercial banks (CBs) in Vietnam, based on data collected from 20 banks during the period from 2013 to the end of 2023 Data were obtained from audited consolidated financial statements and annual reports, published on the official websites of each bank Financial and related indices were calculated by the author to reflect important aspects of bank profitability Relevant macroeconomic data were collected from statistical reports and information published by the General Statistics Office of Vietnam during the aforementioned period

The research employs a multi-stage approach to model estimation and selection In the first stage, Pooled Ordinary Least Squares (Pooled OLS) is employed

7 as a baseline model This approach combines data across all entities and time periods Subsequently, Fixed Effects (FE) and Random Effects (RE) models are estimated to account for potential unobserved heterogeneity that may be specific to individual entities or time periods To determine the most appropriate model for the data, the study utilizes F-tests and the Hausman Test Additionally, the study also employed the Generalized Least Squares (GLS) method to address issues of autocorrelation and heteroscedasticity in the research model.

Research Contribution

From a scientific perspective, the topic contributes to building theoretical foundations and proposing models for research on the relationship between credit risk and profitability of Vietnamese Commercial Banks

From a practical standpoint, the research provides recommendations for mitigating credit risk and suggestions for enhancing operational efficiency to increase the profitability of CBss today As key financial intermediaries, CBs play a crucial role in most economies The efficiency of bank operations can impact economic growth Efficient and profitable operations enable CBs to withstand shocks and contribute to the profitability and financial performance of the financial system

The article focuses on understanding the impact of non-interest income business activities on the operational efficiency of CBs in Vietnam, thereby offering recommendations to improve variables negatively affecting the profitability of these banks This study will provide policymakers and bank managers with empirical evidence of the impact of credit risk on the profitability of CBs in Vietnam from 2013 to 2020 Subsequently, this research will assist policymakers in easily controlling and optimizing the operational efficiency of CBs.

Dissertation Structure

CHAPTER 1: Introduction to the research topic

This chapter provides an overview of the research topic, including the rationale, research questions and objectives, and subjects of the study, research methodology, as well as the novel contributions of the research

Theoretical foundations, this chapter includes the theories of credit risk and profitability of commercial banks and the factors influencing credit risk of commercial banks It briefly introduces some previous studies both domestically and internationally, comparing different viewpoints on the topics Moreover, it summarizes previous research models on credit risk to serve as the basis for proposing the research model

CHAPTER 3: Research model and methodology

This chapter presents the specific methodology used in the study, research design, detailed description of the steps in the research process, as well as data collection and processing methods This section will clarify the regression model used for the study, explaining the variables, their calculations, and expected signs of the variables

CHAPTER 4: Implementation of the research model and analysis

The primary goal is to illustrate the model's results: descriptive statistics for the study sample, model correlation analysis, testing for multicollinearity, testing for heteroscedasticity, and testing for autocorrelation In addition, the feasible generalized least squares (GLS) method is used to solve the difficulties of heteroscedasticity and autocorrelation, which determine the model's final conclusion

Based on the regression model's estimation results with the aforementioned models, the anticipated outcome is to assess how credit risk affects the profitability of commercial banks (CBs), as well as additional significant variables Banks can use

9 this study to gain insights, increase profitability, and reduce credit risk in their operations

This chapter provides an overview of the research topic, the rationale for the topic selection, the research objectives, the research questions, the study's subjects and scope, the research technique, and the data that correspond to each research question The chosen topic involves utilizing statistical software for quantitative analysis, and the results of the model will be compared with previous models to identify factors influencing the operational efficiency of Vietnamese commercial banks

THEORETICAL FOUNDATIONS

Theoretical Background

2.1.1 The concept of Credit Risk

2.1.1.1 Definition of Bank Credit Risk

The credit relationship that exists between commercial banks (CB) and credit institutions (CI) and individuals or businesses (borrowers) is known as bank credit

In this arrangement, assets are temporarily transferred from banks or other lending institutions to debtors Borrowers are required to pay back the credit institutions' principal as well as interest when they mature

Credit risk refers to the potential losses that might occur when borrowers fail to pay back loans in full, including principal and interest, or when they make late payments after receiving credit facilities (both on and off the balance sheet) Credit risk is the likelihood that borrowers or bank counterparties won't stick to the terms of repayment that have been agreed upon, according to the Basel Committee on Banking Supervision (BCBS) Default risk, often known as credit risk arises from uncertainties regarding borrowers' repayment of debts to banks

2.1.1.2 The types of Credit Risk

There are two types of credit risk which is portfolio risk and transaction risk

• Portfolio risk is classified into two categories: intrinsic risk and concentration risk (Nguyen Tuyet Anh, 2021) o Intrinsic risk is derived from the unique characteristics of each borrowing company or economy sector o Concentration risk is the amount of loan exposure concentrated among specific clients, economic sectors, loan types, or geographic areas

• Transaction risk has three components: selection risk, collateral risk, and operational risk o Selection risk is related to the credit assessment and analysis process o Collateral risk stems from collateral standards

12 o Operational risk refers to the management of loan activities

❖ Non-performing loan (NPLs) ratio

The research will utilize the non-performing loan (NPL) ratio to measure the credit risk of the bank This is an important indicator that is easy to collect data on and is clearly recorded on documents and bank systems Non-performing loans are a common phenomenon in today's banking sector The higher the NPL ratio, the greater the Credit Risk Reserve for the Total Deb The risk that customers will default on their loans to the bank is significant, which could lead to capital loss, decreased revenue, and profits

Non-performing loan ratio = Non-performing loans / Credit outstanding

2.1.3 Concept of the Commercial Bank’s Profitability

Credit Risk Measurement Indicators by Nguyen Thi Thu Huyen (2017), profitability is a fundamental indicator for evaluating financial organizations' financial performance, as it takes into account both business results and resource usage Profitability is a critical basis that enables financial institutions to innovate, diversify their products, and operate more efficiently This study evaluates the profitability of financial institutions using the return on assets (ROA) ratio

Before making an investment decision, investors can use the ROA ratio to acquire a enhanced insight of how efficiently a company's assets are utilized This ratio is widely used to evaluate a company's health to that of its competitors in the same industry and broader marketplaces The ROA ratio shows how much profit a company makes for every dollar of total assets it owns A company with a high and consistent ROA in long – term is a good sign that it is becoming more effective in using its assets and optimizing its resources According to international regulations, a company's ROA ratio must be at least 5%.

Empirical Reseach Overview

2.1.4 The impact of Credit Risk on the Profitability of CBs in Vietnam

Credit risk appears when clients fail to repay their loans to the bank (Africa;

2016) The investment portfolio theory is utilized in banking based on the quantity of money the bank has to supply to customers, such as the credit limit granted to third parties Previous research has demonstrated that credit risk has a detrimental influence on bank profitability (Kadri et al 2018), (Hambolu et al 2022) & (Punyata et al 2022) This effect happens when commercial banks are unable to reduce the level of bad loans, resulting in higher bank costs and a decrease in the efficiency of bank operations Credit risk has a detrimental impact on CBs' profitability

Norlina, Bakri & Kelvin (2018) The study focuses on how credit risk affected the profitability of Malaysian commercial banks between 2005 and 2012 It uses non- performing loans (NPLs) as a proxy for credit risk and investigates how banks' credit risk reserves affect profitability Regression analysis uses pooled OLS and panel estimation methods Furthermore, the report concludes that the mortgage crisis had little impact on the Malaysian banking sector due to its limited exposure to US subprime lending products In examining financial system stability and resilience, Beatty and Liao (2009) highlight the importance of loan-loss provisioning policy This strategy plays a vital role in determining swings in banks' profitability and capital levels, which in turn affect their ability to provide credit to the economy

Ahmed, Hambolu, Victor & Abdul (2022) examine the impact of Credit Risk

(CR) on the profitability of eleven Commercial Banks in Nigeria The authors look for data in the sampled banks covering the years 2008–2018 on the correlation between credit risk and Return on Assets (ROA) The authors look for data in the sampled banks covering the years 2008–2018 on the correlation between credit risk

14 and Return on Assets (ROA) The authors then go on to examine the specific implications of every category of credit risk on the profitability of the banking sector in Nigeria According to the authors, this result suggests that if the bank lends money based on the deposit, it will lose money The Capital Adequacy Ratio (CAR) is a novel variable included in this study to evaluate its impact on profitability The results indicate that banks with a good credit quality that comply to credit risk regulations have risk exposure

Shetty and Yadav (2019) investigated the correlation between financial risk and profitability among forty-three Indian commercial banks over 11 years They looked at financial risk indicators including foreign exchange risk (FER) and interest rate risk (IRR), as well as profitability metrics such as return on equity (ROE) and return on assets (ROA) They discovered that whereas ROA was greatly impacted by IRR, ROE showed a limited association with IRR through panel data regression analysis using fixed and random effects models Furthermore, FER had a considerable impact on ROE and ROA

Mohammed, Sani Abdu & Muhammad (2023) investigated the credit risk impact on the value of shareholders in eight Nigerian banks from 2015 to 2019 A panel data regression technique was used to evaluate the data, with non-performing loans (NPLs) as an indicator of credit risk and Market Capitalization (SHV) as variables for the shareholders of the banks under consideration The data found that whilst loan loss provision had a significant negative effect on the value of shareholders, bank loans and advances had a beneficial impact on the increase in the enterprise value of Nigeria's listed banks

(Kithinji, 2010) investigated the influence of credit risk management practices on the profitability of Kenyan commercial banks The analysis encompassed data on total loans, non-performing loans, and total profits for the period 2004-2008 The findings revealed no statistically significant relationship between total loans, non- performing loans, and bank profitability However, this study's analysis demonstrates

15 a limited correlation between commercial bank profitability and the volume of credit issued or the prevalence of non-performing loans These findings suggest that factors beyond credit risk management significantly impact bank profits Consequently, commercial banks seeking to optimize profitability should prioritize strategies that extend beyond traditional loan-centric approaches

Ng Kim Chi, Ly Hoang Khang & L.V Ky Nam (2021) conducted a study to discover factors influencing the profitability of commercial banks in Vietnam They used panel data regression methods, such as Ordinary Least Squares (OLS) and Feasible Generalized Least Squares (FGLS), to evaluate data from 18 Vietnamese commercial banks' financial reports from 2010 to 2019 Based on this analysis, various suggestions were made to individual investors that will assist bank investors and managers in recognizing the detrimental effects of internal and external factors on the profitability and financial stability of these institutions The study found that the following variables influence credit risk: loan portfolio size (LOAN), bank size (SIZE), credit risk (LLR), inflation rate (INF), and GDP

Nguyen Thi My Hanh & Thu (2021) investigated how credit risk management affected the profitability of 30 Vietnamese banks from 2012 to 2020 The study found a significant correlation between asset return, equity return, net interest margin, and risk factors such as capital adequacy ratio, bank size, liquidity ratio, loan to asset ratio, deposit to asset ratio, management efficiency, non-interest income, and GDP growth However, interestingly, the liquidity ratio has a significantly negative relationship with asset return This research provides valuable insights for Vietnamese banks to improve their credit risk management strategies and ultimately boost their profitability

The study of Hung (2016) illustrated the determinants of bank profitability in 30 Vietnamese commercial banks The study found that a bank's capital strength is a crucial factor influencing its profitability Banks with stronger capital are perceived as less risky,

16 which allows them to generate higher returns In particular, retained earnings and valuation adjustments contribute significantly to a bank's capital base On the other hand, the variable

"KCV" (value of customer loans) was found to have a negative impact on profitability This result may contradict findings from international studies but aligns with the reality in Vietnam, suggesting that loan risk management practices in Vietnamese banks were not fully effective during the period from 2008 to 2014 This was particularly evident in 2011, when the bad debt ratio reached alarming levels as Vietnamese banks were burdened by the non- performing loans of state-owned enterprises

Table 2 1 Summary of previous studies

Dependent Variables Independent Variables Model Impacted Result

- Loan to total loan ratio: NPL/

- Loan loss provision to total loan ratio: LLP/LA

- Total deposit ratio LA/TD

Pooled OLS LTD has a positive effect on ROA

- Loan to total deposit: LTD

- Non-performing loans to Total Assets: NPL

- Size : Log of Total Asset

LLP, Size have a positive significant impact on ROA, while

- Risk management derivative product: OBS 2

IRR has a positive significant impact on ROA & ROE

- Non-performing loans to Total Assets: NPL

NPL has a significantly negative on the shareholders’ value

Kenya CBs ROA - Amount of Credit / Total Asset R- square

Total loans and non- performing loans has

- Non – performing loans/ Total loans

- Profits/ Total Assets no significant effect on the profitability

Pooled OLS, FEM, REM and FGLS

Size and GDP have a positive correlation on ROE, white LLR has a negative one

- Loan to asset ratio: LOAN

- Deposit to asset ratio: DEP

Size, CA and ME have a significantly positive correlation on ROE, ROA and NIM

- TG: Value of deposit at SBV and otherint financial intitutions

- KCV: Value of customers and other financial intitutions’ loans

VCSH has a considerately positive on the profitability, while KCV affects negatively

Table 2 2 The state of each national’s finance situation and shortcoming in these previous studies

Author State of the ecnomic finance in research period Shortcoming

During the period from 2005 to 2012, Malaysia experienced only marginal growth of 0.3%, primarily attributable to the lingering effects of the Asian Financial Crisis and the subsequent mortgage crisis These crises were exacerbated by the country's minimal exposure to US subprime loan products

- The independence of variables is too limited, with only the Loan Loss Provision variable being used to measure the credit risk of banks

- The model is only regressed using Pooled OLS, thus its persuasiveness is not high

The Nigerian economy in 2008-2018 struggled with overdependence on oil, leading to vulnerability during price drops This situation had detrimental implications for both the Nigerian economy as a whole and its banking sector

- This analysis solely pertains to microeconomic and internal variables of the bank, with no consideration of macroeconomic factors or the economic situation of each observation

During the period from 2008 to 2018, the Indian economy experienced significant growth, with GDP doubling in 2018 compared to 2008, driven by the development of its economies The Indian economy managed to navigate the global finance crisis well and embarked on a reform path two years after the Global Financial Crisis

- There is a disproportionately low number of independent variables compared to dependent variables

- The analysis fails to account for the factor of financial crises

From 2012 to 2020, the Vietnamese economy remained relatively stable but was impacted by the COVID-19 pandemic, leading to a sharp decline in GDP growth in 2020

- The research period is relatively short

RESEARCH MODEL AND METHODOLOGY

Sampling Technique

The study utilized secondary data obtained from audited annual reports and financial statements of a sample of 20 Vietnamese commercial banks from 2013 to

2023, encompassing the period affected by the COVID-19 pandemic The aim was to investigate the credit management practices of banks during the pandemic An ex- post factor research design was employed, given that the data collection had already transpired, and the researcher lacked control over the events The study population comprised banks listed on the stock exchange Additionally, the author also incorporated data sourced from the World Bank and STB.

Research Model

Table 3 1 Variables used in the research model

Step 1 Define the research question

Step 2 Identify the theories to be employed in the research, determine the estimation model

𝑇𝑜𝑡𝑎𝑙 𝑂𝑢𝑡𝑠𝑡𝑎𝑛𝑑𝑖𝑛𝑔 𝐿𝑜𝑎𝑛𝑠 - Loan to Total Deposit LTD 𝑇𝑜𝑡𝑎𝑙 𝐷𝑒𝑝𝑜𝑠𝑖𝑡 𝑇𝑜𝑡𝑎𝑙 𝐿𝑜𝑎𝑛 +

Bank Size SIZE Log of Total Assets +

GDP Growth GDP The GDP per capita growth rate +

Note: + stands for posive impact, - stands for negative impact

This study used several models that have been used in other studies, such as Ravi (2012) study in Nepal’s commercial banks on the impact of credit risk on financial performance using panel multiple regression

Another model used in this study was that of Nguyen Kim Chi (2021), where the ROE was used as an indicator of profitability, and other proxies such as nonperforming loans were used as indicators of credit risk, and total loan as an indicator of credit risk for banks in Vietnam

The study used a panel regression to determine the causal association between the dependent variable (ROA) and the Independent and control variables are used, such as: LLP (Loan Loss Provision), LEV (Leverage), LTD (Loan to Total Deposit),

NPL (Non-Performing Loans to Total Assets), Size (Log of Total Assets) and GDP

ROA (Return on Asset): According to (My Hanh &Thu; 2021) and

(Hambolu; 2022), this variable is a metric that represents the profitability ratio of bank i's total assets for the year t This ratio shows the bank's profit per unit of assets, which demonstrates how well the bank manages its assets A bank that uses its resources more effectively is indicated by a higher ROA index

LLP (Loan Loss Provision): According to the previous research conducted by the group (Nguyen Kim Chi; 2021), the Loan Loss Provisions variable has an inverse impact on the profitability of the bank (ROA) Specifically, the Loan Loss Provisions variable indicates that higher credit risk needs larger provisions, limiting working capital and, as a result, commercial banks' profitability Excessive credit risk provisioning by the bank indicates a high credit risk coefficient, which suggests poor credit management and the bank's low credit quality and (Ahmad, Tahir, Aziz, & Management, 2014) finding the following evidence that there is a negative relationship between the loan loss provision (LLP) and profitability

Hypothesis 1: LLP (Loan Loss Provision) is expected to have a negative impact on bank profitability

LTD (Loans-to-Deposits Ratio): The total loans to total deposits ratio assesses the extent to which customer deposits are used to fund customer loans (Norlina 2018) This variable may show the bank's use of deposits for profit- generating activities, particularly client lending A higher Loan-to-Deposit Ratio (LTD) suggests a wider scope of activities, resulting in increased profitability (Hambolu, 2022) The research of (Ahmad et al., 2014) also say that there is a negative relationship between deposits and profitability

Hypothesis 2: LTD (Loan-to-Deposit) is expected to have a positive impact on bank profitability

NPL (Non-Performing Loans): Base on the study of (Norlina; 2018),

(Hambolu;2022), the ratio of non-performing loans to total loans serves as a key indicator of a bank's asset quality and directly impacts its profitability Loans are categorized as non-performing when they fall into delinquency stages (typically categories 3 to 5) A consensus exists among researchers that a higher NPL ratio signifies a deterioration in asset quality on the bank's balance sheet This directly translates to lower profitability When issuing loans, banks are obligated to set aside provisions for potential loan defaults The higher the NPL ratio, the larger these provisions need to be, consequently reducing the bank's overall profitability

1 Standard Loans Less then 10 days 0%

2 Watchlist Loans From 10 to 30 days 5%

3 Substandard Loans From 30 to 90 days 20%

4 Doubtful Loans From 90 to 180 days 50%

According to Clause 8 of Article 3 of Circular 11/2021/TT-NHNN

Hypothesis 3: NPL (Non-Performing Loans) is expected to have a negative impact on bank profitability

LEV (Leverage): measured by the debt-to-equity ratio Higher leverage allows banks to potentially amplify returns by utilizing borrowed funds for increased lending activities However, this strategy also introduces greater risk, as a higher debt burden reduces a bank's capacity to absorb potential loan losses Additionally, highly leveraged banks are more susceptible to fluctuations in interest rates Chalermchatvichien & et.al (2014)

Hyphothesis 4: LEV (Leverage) is expected to have a positive impact on bank profitability

Size (Bank Size): Size as measured by the logarithm of total assets, has a convoluted relationship with profitability While larger banks are often more profitable, this trend may not continue permanently According to studies, crossing a specific size threshold might result in inefficiencies and weaker regulations, thus reducing profitability This suggests a nonlinear link between size and profitability

In contrast, empirical research by Staikouras et al (2004), (Ahmad et al., 2014) and Dietrich et al (2011) shows a positive relationship between bank size and profitability for commercial banks

Hypothesis 5: Size (Bank Size) is expected to have a positive impact on bank profitability

GDP (Gross Domestic Product Growth): While the World Bank compiles data on annual GDP growth, the relationship between this growth and bank profitability (measured by Return on Assets or ROA) remains unclear Some studies, like (Bhattarai & Research, 2016), find no statistically significant impact of GDP growth on ROA However, other research presents contrasting findings Bertay et al

(2018) reveal a positive and substantial correlation, suggesting that higher GDP growth leads to improved bank profitability Conversely, (Pan & Pan, 2014)

25 demonstrates a significant negative impact, indicating that economic expansion might hinder bank profitability

Hypothesis 6: GDP (Gross Domestic Product) is expected to have a positive impact on bank profitability.

Research Data

The study is based on imbalanced panel data gathered from the financial and annual reports of 30 commercial banks from 2013 to 2023, as well as the State Bank of Vietnam website Additionally, macroeconomic data were obtained from the World Bank and the General Statistics Office of Vietnam.

Regression methods and testing

The relationship between credit risk and bank profitability is estimated in the study paper using the FEM, REM, and Pooled OLS methodologies The existence of autocorrelation and heteroskedasticity in the model residuals is then evaluated using the White heteroskedasticity test and the Wooldridge test In the event that these tests reveal the existence of any problem, the GLS technique is used to produce more accurate and efficient estimates

3.5.1 Pooled Ordinary Least Squares (Pooled OLS)

All of the coefficients in the Pooled Ordinary Least Squares (Pooled OLS) model is a basic regression analysis technique that assumes that all observations are drawn from the same population and that there are no unobserved individual effects stay consistent across time and amongst people Because it ignores the temporal and spatial components of panel data, this is the most basic and uncomplicated approach Rather, it use a conventional OLS regression to estimate coefficients

The Fixed Effects Model (FEM) is a regression analysis technique that is especially useful for evaluating panel data, which is made up of observations from several entities or individuals over time FEM assumes that each people or entity has

26 unique traits, called as fixed effects, that might impact the outcome variable These fixed effects are usually unnoticed and cannot be readily measured

The general form of the FEM equation is:

• Yit is the value of the outcome variable for individual i at time t

• Ci is the fixed effect for individual i

• β is the coefficient for the independent variable Xit

• Uit is the error term for individual i at time t

The Random Effects Model (REM) is another useful regression analysis tool for panel data Unlike the Fixed Effects Model (FEM), which posits that each individual or entity has distinct fixed effects, the REM assumes that unobserved individual effects are random and unrelated to the independent variables This indicates that the REM's regression equation does not include dummy variables for each individual or entity The general form of the REM equation is:

• Yit is the value of the outcome variable for individual i at time t

• β is the coefficient for the independent variable Xit

• εi is the random effects term for individual i

• uit is the error term for individual i at time t

While OLS estimation for the random effects model produces unbiased parameter estimates, it comes at the cost of inefficiency This inefficiency stems from OLS's disregard for the autocorrelation present within the error component μit To achieve both unbiasedness and efficiency in the estimates, Generalized Least Squares (GLS) estimation can be utilized GLS effectively addresses the problems of serially correlated and heteroscedastic errors within the chosen model, as confirmed by the Hausman test

Implementation Method of Research Model

Chapter 3 expands on the theoretical framework and research presented in Chapter 2 by proposing research hypotheses about the internal and external factors that influence the relationship between credit risk and profitability in Vietnamese commercial banks

In this chapter, the author thoroughly defines the research technique, creates the research model, and combines many factors into it The chapter also describes the phases in the research process, such as descriptive statistics, model selection, testing, and other procedures for determining the best model

The proposed research model is explicitly described as follows:

IMPLEMENTATION OF THE RESEARCH MODEL AND

Research Results of the Model

4.1.1 Descriptive statistics of the data

To understand the fundamental characteristics of the research variables, such as the maximum value, minimum value, mean value, and the deviation between the mean values of the variables and the actual values, it is essential to conduct a descriptive statistical analysis of the research sample The results of the descriptive statistics are presented in Table 4.1 below:

Table 4 1 Descriptive statistics of the data

Variable Obs Mean Std dev Min Max

Source: Extraction of results from Stata 17 software

Table 4.1 presents the statistical values for all eight research variables in the model, based on a sample size of 220 observations The ROA (Return on Assets) variable, measuring profitability, ranges from a minimum of -0.0070 to a maximum of 0.2325, with a sample mean of 0.0107 and a standard deviation of 0.0167 Typically, a healthy bank achieves an ROA between 1% and 2% The average ROA of 1.07 % for the banks studied from 2013 to 2023 indicates suboptimal asset management and income production, suggesting caution against excessively high returns due to associated risks

Profitability variations are unavoidable in a rising economy, especially in the banking sector, where cash is the primary business product and is quite liquid As a result, provisioning for risks, particularly credit risks, is critical The large range of data, from 0.0000 to 0.4558, demonstrates significant variability in loan loss reserves This variety could be related to differences in loan credit quality, borrower company sectors, or other risk considerations The average score of 0.0146 indicates a low amount of provisioning for potentially hazardous loans This low average may indicate the banks' conservative approach to credit risk assessment or the high quality of their loan assets Overall, adequate Loan Loss Provisions (LLP) are critical for banks to reduce possible losses and maintain financial stability

The loan-to-deposit (LTD) ratio, which measures the proportion of total loans to total customer deposits, is an average of 88.32%, suggesting that most banks invest up to 90% of their deposits in loans The minimum percentage of 2.55% illustrates that some banks have a more conservative lending policy This high average LTD ratio reflects the fact that many banks use nearly all of their customer deposits for lending, which is their principal source of income With LTD values ranging from 2.55% to 205%, banks have the ability to make significant profits from lending activities However, this aggressive lending strategy entails considerable risks, including liquidity risk, credit risk, and diminished capital efficiency if diversification into other profitable channels is not feasible As a result, while the LTD ratio is positively correlated with bank profitability, an overall assessment of a bank's performance must take into account other independent variables in order to provide a thorough review of operational efficiency.

The Non-Performing Loan (NPL) ratio is a crucial indicator of a bank's loan quality and debt recovery ability It measures the proportion of non-performing loans to total loans In this study, the average NPL ratio is 0.0259, indicating relatively low

32 levels of bad loans among banks However, the large disparity between the maximum value of 0.4369 and the minimum value of 0.0000 shows significant differences in credit risk The standard deviation of 0.0453 further highlights this variation These differences may stem from varied lending strategies, borrower quality, or other risk factors Therefore, commercial banks must closely monitor and manage their loan portfolios to maintain profitability and stability, ensuring robust credit risk management to mitigate potential losses and support financial performance

The leverage ratio (LEV), also known as the debt-to-equity ratio, measures how much money a bank utilizes compared to its capital Commercial banks have relatively low leverage, as seen by an average value of 0.6518 (65.18%) and a standard deviation of 24.9046 This prudence in using borrowed money is intended to reduce risk The standard deviation also shows a large variance in leverage usage between institutions Some banks employ significantly more leverage than the average, whereas others use far less The leverage ratio stays steady, demonstrating that banks are effectively using leverage to increase profits by investing in projects with returns that outperform borrowing costs

Bank size (SIZE) is calculated using the natural logarithm of total assets, which ranges from a minimum of 30.3178 to a maximum of 35.3721, with a sample mean of 32.9237 The standard deviation of SIZE is 1.1462, indicating a significant level of data dispersion Overall, the Vietnamese banking system is marked by a trend of strong capital banks assisting weak capital institutions, with a clear divide between

"rich" and "poor" banks within the same Vietnamese commercial banking system

Figure 4 1 Asset Growth Rate of the 20 Vietnamese Commercial Banks Over the Years

From 2013 to 2023, the total assets of 20 commercial banks increased by 0.192 percent Large banks including BIDV, Vietcombank, Agribank, and Vietinbank saw comparatively significant asset growth Smaller banks, such as KienlongBank and OceanBank, remained relatively stable and did not grow significantly Overall, Vietnam's commercial banking sector is shifting towards stronger-capitalized banks that assist those with lower capital This subject demands additional research to promote mergers and acquisitions (M&A) for integration and development

Gross Domestic Product Growth (GDP) is a crucial indicator of economic health, influencing bank profitability Economic growth boosts loan demand from businesses and individuals, leading to higher bank profits A strong economy also lowers default rates, reducing risk for banks and enabling them to charge higher interest rates or fees, further increasing profitability Our findings are expected to align with previous research, which has shown positive links between economic growth and bank profitability (Zampara, Giannopoulos, & Koufopoulos, 2017) The average economic growth rate is 0.78043, with the highest value at 1.1296 and the

34 lowest at 0.4085 Overall, the variation in economic growth between years is not significant, with a standard deviation of 0.1781 According to Figure 4.2, the GDP growth rate fluctuated during the period from 2013 to 2023 However, following the COVID-19 pandemic, Vietnam's economy experienced a sharp decline in 2020 but showed signs of recovery in 2023, with GDP growing by 8% from 2020 to 2023

Figure 4 2 Gross Domestic Product (GDP) and GDP Growth Rate of Vietnam over the sampling years 4.1.2 Correlation Analysis

Table 4 2 Table of correlation coefficients

ROA LLP LTD NPL LEV SIZE GDP

Source: Extraction of results from Stata 17 software

According to Table 4.2, only one independent factors, NPL and LEV, have a negative relationship with the dependent variable ROA, whereas all other variables have a positive association Among these variables, LTD has a relatively large association with ROA (correlation coefficient of 0.7575), whereas GDP has the least negative correlation with ROA (correlation coefficient of -0.0349) Furthermore, the variables in the model have relatively modest correlations with one another (the majority of the absolute values of the correlation coefficients range from -0.0271 to 0.7575), with no correlation coefficients exceeding 0.8 As a result, the independent variables do not exhibit significant multicollinearity If the correlation coefficients between variables in the correlation matrix were high (more than 0.8), it would imply a strong correlation among those variables, which could lead to model flaws However, because none of the matrix's coefficients surpass 0.8, the ROA model has no multicollinearity difficulties

4.1.3 Pooled Ordinary Least Squares (Pooled OLS) Regression

Source: Extraction of results from Stata 17 software

The OLS regression results in Table 4.3 indicate that the F-test value for the model (Prob > F = 0) is below the 5% significance level, suggesting that the model used for ROA is appropriate Regarding the impact of the variables on the model, the independent variables account for 65.19% of the variation in ROA (R-squared 0.6519) The relatively high R-squared value (greater than 50%) indicates a strong relationship between the independent variables and the dependent variable (ROA).

Model Selection

4.2.1 F-test to Choose Between the Pooled OLS Model and the Fixed Effects Model (FEM)

• H0: The Pooled-OLS model is appropriate for the research sample

• H1: The FEM is appropriate for the research sample

Source: Extraction of results from Stata 17 software

The test results show that the P-value < α = 0.05, providing sufficient evidence to reject the null hypothesis (H0) Therefore, the FEM (Fixed Effects Model) is more appropriate than the Pooled-OLS model

4.2.2 Hausman Test to Choose Between the FEM and the REM

The study uses the Hausman test to evaluate whether the FEM (Fixed Effects Model) or the REM (Random Effects Model) is more appropriate for the research sample

• H0: The REM is more appropriate for the research sample than the

• H1: The FEM is more appropriate for the research sample than the

Table 4 5 The Hausman – test result

Source: Extraction of results from Stata 17 software

The test results show that the P-value < α = 0.05, we have sufficient evidence to reject the null hypothesis (H0) Therefore, the FEM (Random Effects Model) is more appropriate than the REM

Table 4 6 Estimation Results using Pooled OLS, FEM, and REM Methods

Variables Pooled OLS FEM REM

Coefficient P – Value Coefficient P – Value Coefficient P – Value

Selecting Pooled OLS and FEM FEM and REM

There are no differences between the subjects or

There is no correlation between the idiosyncratic errors across subjects and the explanatory variables.

The Pooled-OLS model is appropriate for the research sample

The Fixed Effects Model (FEM) is more appropriate for the research sample than the Random Effects

Source: Extraction of results from Stata 17 software

Conclusion: The results of the OLS, FEM, and REM regressions presented in

Table 4.6 indicate that among the three models (OLS, FEM, and REM) when regressing panel data, the REM model is the most suitable for the overall research data.

FEM Result

Source: Extraction of results from Stata 17 software

The estimated regression model yields a P-value = 0, which is less than 0.05, indicating that the model is suitable The R-squared value is 0.5011, meaning that the independent variables in the model explain 50.11% of the variance in the ROA variable, representing a moderate level of influence

Heteroscedasticity occurs when the residuals or errors (e) of a regression model do not follow a random distribution and have unequal variances after the regression process This violates the assumption of the linear regression model that the variances of the errors should be constant and equal

The author employed the LM-Breusch and Pagan Multiplier test to examine the heteroscedasticity of the model's error variance, assuming the following hypotheses:

• H0: Error variance is homoscedastic (constant across entities)

• H1: Error variance is heteroscedastic (varies across entities)

Table 4 7 The LM-Breusch and Pagan Multiplier test

Source: Extraction of results from Stata 17 software

The test results show that the P-value = 0.000, which is less than α=0.05, providing enough evidence to reject the null hypothesis H0 Therefore, it suggests that there is heteroscedasticity across entities or the FEM model has a heteroscedasticity problem

Autocorrelation occurs when the error values at one point in time are related to the error values at a previous point in time Violating the assumption of independent errors can affect the accuracy and efficiency of the regression model

The author uses the Wooldridge test to check for autocorrelation among the variables, with the following hypotheses:

Source: Extraction of results from Stata 17 software

The test results show that the P-value is less than α=0.05, providing sufficient evidence to reject the null hypothesis H0 Therefore, the FEM model exhibits autocorrelation issues

The author uses the Collinearity Diagnostics tool to test for multicollinearity, with the following results:

Table 4 9 The Collinearity Diagnostic test Variable VIF 1/VIF

Source: Extraction of results from Stata 17 software

The independent and dependent variables all have VIF values at low levels and below 10, indicating that the assumption of no multicollinearity is not violated Therefore, the model does not exhibit multicollinearity issues

Table 4 10 Test Results for the REM Model

Multicollinearity Test Mean VIF = 1.12 No

Note: Yes and No respectively indicate the presence and absence of defects

Source: Extraction of results from Stata 17 software

Table 4.10 summarizes the tests performed on the FEM model The P-value < 0.05 indicates that the FEM model has heteroscedasticity concerns The F-test gives a p-value < 0.05, indicating that the model has autocorrelation problems

Table 4.10 also demonstrates that the research model has no multicollinearity concerns However, the model is flawed in terms of both autocorrelation and heteroscedasticity This will make the estimations from standard regression methods on panel data ineffective and the tests incorrect To address these concerns, the author adopts the the Generalized Least Squares (GLS) estimation approach, which ensures robust and efficient estimates (Wooldridge, 2002).

Mitigating Deficiencies with GLS Model

After applying the aforementioned tests to the FEM model, the findings revealed the presence of heteroscedasticity and autocorrelation in the error variance

As a result, to solve these inadequacies, the author chose to use the Generalized Least Squares (GLS) regression approach The results are presented as follows

Table 4 11 Regression Results using GLS ROA Coefficient Std err z P>z

Source: Extraction of results from Stata 17 software

Table 4.11 presents the regression results of the model explaining the impact of credit risks on the profitability of Vietnamese banks measured through the ROA indicator using the GLS method The estimated model is as follows:

The GLS estimation results show that the regression coefficients of LLP, LTD, NPL, Size, and Constant are significant at the 1% level

The variables LEV and GDP do not have statistical significance in explaining the change in ROA due to their p-values > 5%.

Stability and Effectiveness of the Model

To ensure the stability and effectiveness of the model, the estimates must revolve around the true value and have a small variance in each estimation Specifically as follows:

Table 4 12 Summary of the Intercept Coefficients of the Regression Models Variable Pooled OLS FEM REM GLS

Source: Calculated from the data of the study

Table 4 13 Summary of the Standard Deviations of the Regression Models Variable Pooled OLS FEM REM GLS

Source: Calculated from the data of the study

The results reveal that the Pooled OLS and GLS models generate comparable results for all independent variables (with the exception of GDP) In contrast, the FEM and REM models produce distinct outcomes This is related to model flaws like heteroscedasticity and autocorrelation The GLS model is proposed to address the inadequacies of the FEM model, and the model is resilient when compared to Pooled OLS, with similar outcomes.

Research Discussion

The model that works best is the generalized least squares (GLS) regression model, according to the findings of model selection As a result, the outcomes vary from the initial projections in the following ways:

Table 4 14 Compare the research hypothesis with the regression results Variables Expected Sign Research Result Std Err P > z

Note: + stands for posive impact, - stands for negative impact

As the result, the whole model suggest that LLP, LTD, NPL and Size are significant variables while GDP is not significant variables

LLP (Loan Loss Provision): Based on a prior investigation by Fernando and

Ekanayake (2015), they explored whether commercial banks in Sri Lanka used LLP to smooth income between 2003 and 2012 They identified a positive relationship between loan loss provisions and profits before taxes, implying that banks do not use LLPs to manage their income According to Hypothesis 1, there is a negative relationship between loan loss provision and profitability However, the research findings show an inverse effect with coefficient -0.0111 at a significance level 5% This can be interpreted as commercial banks using LLP and security gain realization to eliminate minor earnings declines By boosting provisions, banks can cut present earnings while potentially enhancing future profits when provisions are reversed Furthermore, a bank with effective credit risk management and increased provisions for poor loans can boost investor and customer confidence while improving operational efficiency and long-term profitability

LTD (Loans-to-Deposits Ratio): The loan-to-deposit ratio's impact on profitability shows a positive coefficient 0.0076 at a significance level lower than 5%, consistent with the early projected sign Commercial banks' profitability does not

45 significantly correlate with their loan-to-deposit ratio It can be explained that a high LTD ratio means that the bank is lending money well, increasing revenue and ROA Since judicious credit expansion usually leads to improved earnings and maintains the bank's operational effectiveness (Hambolu et al, 2022) Banks, as unique businesses dealing with currency, view the loan-to-deposit ratio as having a direct impact on profitability because it represents the utilization of working capital By managing credit risk effectively, banks can maintain high profitability even with a high LTD ratio, as this does not necessarily lead to increased credit risk Additionally, a bank that achieves high profitability while maintaining a high LTD ratio can enhance investor confidence, thereby attracting more capital and facilitating further loan growth

NPL (Non-Performing Loans): The results indicate that NPL has a negative but insignificant impact on ROA, with a regression coefficient of -0.0192 at the 5% significance level, supporting the initial hypothesis 3 This finding is consistent with a previous study by (Gelos, 2006) which stated that NPL has a significant negative effect on ROA This means that NPL reflects credit risk: the lower the NPL, the lower the credit risk borne by the bank This suggests to the author that although an increase in NPL can reduce ROA, with effective management measures and business strategies, this impact can be mitigated by increasing loan loss provisions (LLP) There is a clear inverse relationship between NPL and LLP: as the NPL ratio increases, banks need to increase LLP to cover potential losses from bad debts, thus protecting the bank from potential credit risks

SIZE (Bank Size): The bank size variable has a statistically significant positive correlation with the dependent variable ROA (0.004) at a level lower than 5% This finding is consistent with research by Nguyen Kim Chi (2021) and Trung

(2020), which found that increasing assets helps banks improve opportunities in company operations It also demonstrates that asset size has a beneficial impact on bank profitability, as expansion in size is frequently associated with larger

46 organizational management and control networks The expansion and growth of bank size does have an impact on profitability because it allows access to a wider pool of potential clients Specifically, a 1% increase in bank size (Size) results in a 0.4% rise in ROA

The author conducted a regression analysis using three models: Pooled OLS, FEM, and REM Based on model selection tests, the FEM model was found to be the most appropriate However, upon examining model defects, the FEM model exhibited heteroscedasticity and autocorrelation

To address these shortcomings, the GLS regression method was employed, and the regression results are as follows:

ROAi,t = - 0.0766 + 0.2858 LLP i,t + 0.0069 LTD i,t - 0.0356 NPL i,t + 0.0897 LEV i,t + 0.0024 SIZE i,t - 0.0003 GDP i,t + εi ,t

The final research findings show that the FGLS model has five statistically significant variables at the 1% and 5% levels These include loan loss provision (LLP), loan-to-deposit (LTD), non-performing loans (NPL), and bank size (SIZE) And as the result, the whole model suggest that LLP, LTD, NPL and Size are significant variables while GDP and LEV is not significant variables

In Chapter 5 of the thesis, the author will make recommendations based on the research findings to improve the operating efficiency of Vietnamese commercial banks

CONCLUSION AND RECOMMENDATIONS

Conclusion

The research investigates the impact of credit risk on the profitability of Vietnamese commercial banks through a quantitative model By reviewing previous studies, the research selects several potential variables that have been used and conducts empirical research to examine the effects of these variables on the profitability of banks, measured by the ROA indicator Data was collected from 20 Vietnamese commercial banks over the period from 2013 to 2023 The study employs panel data regression models, including the Pooled OLS, Fixed Effects Model (FEM), and Random Effects Model (REM), and conducts tests to select the most appropriate model and check for common model defects Subsequently, the study uses the Generalized Least Squares (GLS) regression method to address these defects, ensuring the feasibility and effectiveness of the research model

The regression results indicate several factors significantly influence the profitability (ROA) of Vietnamese commercial banks Notably, the loan loss provision (LLP), loan-to-deposit ratio (LTD), non-performing loans (NPL), and bank size (SIZE) were found to be statistically significant variables In contrast, GDP growth and LEV rate were not statistically significant, suggesting that these factors do not have a direct impact on banks' profitability in the short term

The negative impact of the non-performing loans (NPL) ratio on the profitability (ROA) of commercial banks highlights the importance of managing credit risk effectively The increase in loan loss provision (LLP) reflects a proactive approach to setting aside reserves to cover potential loan losses, indicating strong risk management practices Additionally, a lower level of non-performing loans (NPL) suggests that banks are effectively managing their credit risk Consequently, an increase in leverage, supported by these strong risk management practices, leads to an increase in profitability (ROA)

The insignificance of the GDP growth rate in influencing ROA may indicate that macroeconomic conditions, as measured by GDP, do not directly translate to changes in bank profitability This could be due to banks' operations being more sensitive to other factors such as credit risk management and internal efficiencies than broad economic performance.

Recommendations

Firstly, to efficiently manage and settle non-performing debts, Vietnam's commercial banks should take a diversified approach First, they should prioritize direct debt collection and collateral asset liquidation in accordance with regular debt categorization results Banks can speed up the process of collecting bad debts by assessing and developing customized recovery plans Second, debt restructuring should be used when debtors are experiencing temporary financial difficulties but have the capacity to repay if given longer terms Third, banks can relieve financial pressure on borrowers by lowering or eliminating interest, encouraging them to repay a portion or all of their outstanding loan Fourth, debt trading with expert debt purchase firms can quickly recoup some of the bank's cash, allowing it to reinvest in new prospects Fifth, when alternative techniques fail to produce results, legal action should be taken as a last resort Finally, banks should consider alternative options, such as corporate reorganization and the creation of specialized departments, to proactively monitor and address developing bad debts, assuring a comprehensive and successful bad debt resolution approach

Secondly, diversification in Banking Activities is crucial key With rising rivalry in the banking sector, compared to global banks, there are other concurrent activity, particularly in services Domestic commercial banks (NHTM) continue to rely heavily on credit-related profits To grow sustainably in today's competitive and risky banking market, commercial banks must diversify and innovate in their business offerings Diversification of activities will help to reduce risks compared to traditional and heavily weighted credit activities Relying primarily on lending and

50 borrowing can be extremely dangerous during economic downturns Furthermore, as competition in the money market heats up, the quality and variety of services will provide a competitive advantage As previously established, boosting nontraditional business activity improves banking efficiency Therefore, expanding into other industries, particularly service-related operations, is critical in the current environment Banks should invest in and provide recommendations to build modern banking services based on the type of service, whether for individuals or enterprises, such as fund management, safe deposit box rentals, financial transactions, and so on

At the same time, accelerate and modernize information technology systems and infrastructure to guarantee that customer transactions are secure, quick, and easy This will lay the groundwork for the implementation and development of more diverse and high-quality technology-driven products and services

Thirdly, increasing equity capital and developing operational scale are critical steps toward maintaining commercial banks' business stability in Vietnam This technique promotes loan growth and improves the net interest income ratio To accomplish this, banks must actively mobilize cash from a variety of sources, expand their branch networks, and invest in current technology Furthermore, adhering to the State Bank of Vietnam's mandated reserve requirements and routinely provisioning for credit risk are critical strategies to mitigate liquidity risks Banks must comply with mandated reserve ratios, make proper provisions for credit risk, implement early warning systems, and regularly assess risks Furthermore, boosting credit risk management quality through staff training, implementing scientific credit scoring systems, and strengthening internal oversight and auditing is critical to reducing bad debt ratios and increasing credit management efficiency These initiatives will assist Vietnam's commercial banks in improving their risk management capabilities, maintaining stability, and achieving long-term growth in a dynamic economy

Lastly, after the COVID-19 epidemic ended in 2021, commercial banks increased their provisions for bad debt in group 5 Although this technique improves

51 risk resilience and reliability, it is transient and may result in significant provisioning expenses in the long run, raising liquidity risk As a result, bank managers must use flexible economic methods to keep provisioning at a moderate level, ensuring that the bank has enough risk reserves without incurring a major liquidity cost burden in the future.

Research Limitations

With the aforementioned topic, the author draws conclusions based on data obtained from 20 Vietnamese commercial banks over a 10-year period, from 2013 to 2023, yielding a relatively modest number of findings As a result, the study's findings are influenced by the economic conditions of the time period, the number of domestic commercial banks, the data published by these banks each year, and the number of institutions investigated The conclusions drawn are only valid within the selected timeframe, limiting their spatial breadth

Besides, the author's findings are based on publicly available data from the analyzed commercial banks In other words, the study is constrained by the accuracy of its data

Moreover, the author's research is conducted with the knowledge level of a student obtaining a degree in economics, who may lack the most precise assessment and calculating methodologies for the issues at hand The dependent variable is confined to ROA, however other variables like as ROE, NIM, and EPS could be used to quantify bank profitability

Furthemore, the study focuses on a number of issues, including bank size, operational costs, financial leverage, GDP, and inflation However, in reality, numerous other factors influence bank profitability, including interest rates, funding structure, bad debt, and foreign investor holdings As a result, the study may not fully address all of the issues raised Finally, the study focuses on Vietnamese joint-stock commercial banks.

Proposed Directions for Further Research

Based on the aforementioned limitations, the author proposes the following directions for further research to deepen the investigation on this topic

Firstly, to enhance the objectivity of the evaluation results, the author suggests supplementing the missing data Regarding data collection, it is proposed to extend

52 the study period Eleven years may not be sufficient to draw conclusions; hence, extending it to 20 years could provide more reliable results Simultaneously, expanding the geographical scope of the research is recommended Besides the 20 selected companies studied by the author, other companies from various sectors should also be chosen to validate the proposed hypotheses Additionally, conducting detailed studies by specific regions such as North-Central-South or by economic key areas would contribute to a more accurate research topic

Secondly, in the data processing stage, the author proposes using the GMM model for analyzing panel data with the goal of addressing the shortcomings encountered when estimating FEM and REM models

Thirdly, considering additional variables such as operating costs with ROA or

ROE to gain an overall view of cash flows, thereby evaluating the profitability of companies

According to the experimental research presented in Chapter 5, the study made numerous recommendations to improve the operational efficiency of commercial banks in Vietnam These ideas include measures to resolve non-performing loans, increase scale and financing sources, plan for risk management, and diversify business operations The report noted that enhancing bank profitability necessitates efforts from management and employees and ongoing support from the government and the State Bank of Vietnam Recognizing limitations such as the short study period (2013-2023) and a sample size of 20 banks, the author proposed extending the study period to 20 years, including more banks for better data accuracy, and employing the GMM model for panel data analysis to improve on the limitations of FEM and REM models These steps are aimed at assisting Vietnamese commmercial bank managers in formulating income diversification policies to enhance operational efficiency and achieve their goals effectively

Ahmad, F., Tahir, S H., Aziz, B J T s R J o S S., & Management (2014) Impact of loan loss provision on bank profitability in Pakistan 3(12), 34-41

Alhadab, M., Alsahawneh, S J I J o B., & Management (2016) Loan loss provision and the profitability of commercial banks: Evidence from Jordan 11(12), 242-248

Al-Homaidi, E A., Tabash, M I., Farhan, N H., & Almaqtari, F A

(2018) Bank-specific and macro-economic determinants of profitability of Indian commercial banks: A panel data approach.Cogent Economics & Finance,6(1),

Anh, & Nam, N Nâng cao năng lực tài chính của ngân hàng thương mại:

Nghiên cứu điển hình tại Ngân hàng

Bala, S A R., Auwal, B M., Salisu, M Y J I J o A., Finance, & Management, R (2023) The Impact of Credit Risk on the Value of Shareholders of Listed Banks in Nigeria 7(1), 27-33

Bhatt, T K., Ahmed, Naveed Iqbal, Muhammad Babar Ullah, & Mehfooz

(2023) Examining the determinants of credit risk management and their relationship with the performance of commercial banks in Nepal 16(4), 235

Bhattarai, Y R J P I J o M., & Research (2016) Effect of non-performing loan on the profitability of commercial banks in Nepal 10(2), 1-9

Bordeleau, É., & Graham, C (2010) The impact of liquidity on bank profitability

Butola, Punyata Dube, Praveen Jain, & Kumar, V (2023) A study on impact of credit risk management on the profitability of Indian banks 11(3), 103-114

Dietrich, A., & Wanzenried, G (2009, April) What determines the profitability of commercial banks? New evidence from Switzerland In12th conference of the Swiss society for financial market researches, Geneva(pp 2-39)

Gelos, M G (2006) Banking Spreads in Latin America (EPub): International Monetary Fund

Geller, Stacie E Koch, Abigail R Garland, Caitlin E MacDonald, E Jane Storey,

& Francesca Lawton (2018) A global view of severe maternal morbidity: moving beyond maternal mortality 15, 31-43

Hambolu, Victor Olufunsho Omuemu, Stanley Osasere Majeed, & Abdul, A

T (2022) The impact of credit risk on the profitability of commercial banks in Nigeria 3(2)

Huỳnh Thanh Tuấn (2020), Các yếu tốtác động đến khảnăng sinh lời của ngân hàng thương mại tại Việt Nam, Luận văn Thạc Sĩ, Trường Đại Học Ngân Hàng TP

Karim, Bakri Abdul Kadri, Norlina Min, & Yong, K L (2018) Credit Risk and Banks’ Profitability in Malaysia 2(1)

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

Koch, Timothy W MacDonald, Steven Scott He, & Ziyun (2000) Bank management: Dryden Press Orlando

Laeven, L & Levine, R (2009), ‘Bank governance, regulation and risk taking’, Journal of financial economics, 93(2), 259-275

Lý Hoàng Khang, Lê Vũ Kỳ Nam, & Chi, N K (2021) Các yếu tố tác động đến khả năng sinh lời của các Ngân hàng Thương Mại Việt Nam (Bachelor), Trường Đại học Công nghiệp Thành phố Hồ Chí Minh, Hội nghị Khoa học trẻ lần 3 năm

Nguyễn Thị Hồng Ánh, & Trung, L T (2020) Tác động của rủi ro tín dụng đến sự ổn định tài chính của các ngân hàng thương mại Việt Nam Tạp chí Công

Nguyễn Tuyết Anh (2021) Rủi ro tín dụng là gì? Cách phân loại rủi ro tín dụng Luanvan1080

Norlina Kadri, Bakri Abdul Karim, & Ming, K L Y (2018) Credit Risk and Banks’ Profitability in Malaysia Universiti Malaysia Sarawak (UNIMAS)

Pan, Q., & Pan, M J O J o S S (2014) The impact of macro factors on the profitability of china’s commercial banks in the decade after WTO Accession 2(9), 64-69

Poudel, R P S J I J o a., & commerce (2012) The impact of credit risk management on financial performance of commercial banks in Nepal 1(5), 9-15

Saleh, D S., Winarso, E J I J o M., & Understanding, M (2021) Analysis of non-performing loans (NPL) and loan to deposit ratio (LDR) towards profitability 8(1), 423-436

Sari, L., Septiano, R J J o A., & Management, F (2020) Effects of intervening loan to deposit ratio on profitability 1(5), 228-241

Shetty, Chetan Yadav, & S, A (2019) Impact of financial risks on the profitability of commercial banks in India 7, 550

Supervision, & Banking (2011) Basel committee on banking supervision

Tan, Y (2016) The impacts of risk and competition on bank profitability in China.Journal of International Financial Markets, Institutions and Money,40, 85-110

Trung, L Đ D (2020, 12/2020) Các nhân tố tác động tới khả năng sinh lời của ngân hàng thương mại tại Việt Nam: Tiếp cận theo mô hình thực nghiệm động

Tạp chí Công Thương Retrieved from https://tapchinganhang.gov.vn/cac-nhan-to- tac-dong-toi-kha-nang-sinh-loi-cua-ngan-hang-thuong-mai-tai-viet-nam-tiep-can- theo-mo-hi.htm

Wooldridge, J M J C., ma (2002) Econometric analysis of cross section and panel data MIT press 108(2), 245-254

Yao, H., Haris, M., & Tariq, G (2018) Profitability determinants of financial institutions: evidence from banks in Pakistan International Journal of Financial Studies,6(2), 53

57 Zulfikar, A (2018) Disclosure of isr and gri index on sharia banking in indonesia Nisbah: Jurnal Perbankan Syariah,4(1), 64-71

APPPENDICES APPENDIX 1 LIST OF COMMERCIAL BANK IN RESEARCH

No Brand Name Registered English Name Time

1 ACB Asia Commercial Joint Stock Bank 2013 - 2023

2 ABBank An Binh Commercial Joint Stock Bank 2013 - 2023

3 BIDV Bank for Investment and Development of

4 Vietinbank Vietnam Bank for Industry and Trade 2013 - 2023

5 Eximbank Vietnam Export-Import Commercial Joint

6 HDBank Ho Chi Minh City Development

7 Kienlongbank Kien Long Commercial Joint Stock Bank 2013 - 2023

8 MB Military Commercial Joint Stock Bank 2013 - 2023

9 MSB Vietnam Maritime Commercial Join Stock

10 Nam A Bank Nam A Commercial Joint Stock Bank 2013 - 2023

11 NVB National Citizen Commercial Joint Stock

12 OCB Orient Commercial Joint Stock Bank 2013 - 2023

13 SeABank Southeast Asia Commercial Joint Stock

14 SAIGONBANK Saigon Bank for Industry and Trade 2013 - 2023

15 SHB Saigon-Hanoi Commercial Joint Stock

16 Sacombank Sai Gon Thuong Tin Commercial Joint- stock Bank 2013 - 2023

17 Techcombank Viet Nam Technological and Commercial

18 Vietcombank Joint Stock Commercial Bank for Foreign

19 VIB Vietnam International Commercial Joint

20 VPBank Vietnam Prosperity Joint Stock

Bank Year ROA LTD NPL LEV SIZE LIQ GDP

ABB 2013 0,0024 0,6363 0,0431 0,1107 31,6850 0,3114 5,5535 ABB 2014 0,0017 0,5758 0,0481 0,0926 31,8426 0,3069 6,4222 ABB 2015 0,0014 0,6504 0,0212 0,0988 31,7957 0,2160 6,9872 ABB 2016 0,0033 0,7724 0,0213 0,0855 31,9374 0,1961 6,6900 ABB 2017 0,0058 0,8274 0,0277 0,0781 32,0678 0,1851 6,9402 ABB 2018 0,0079 0,8382 0,0189 0,0826 32,1308 0,1352 7,4650 ABB 2019 0,0098 0,8164 0,0231 0,0828 32,2614 0,2478 7,3593 ABB 2020 0,0096 0,8729 0,0209 0,0829 32,3878 0,2581 2,8654 ABB 2021 0,0129 1,0169 0,0234 0,1074 32,4263 0,2174 2,5616 ABB 2022 0,0104 0,9749 0,0288 0,1115 32,4997 0,1820 8,0198 ABB 2023 0,0028 0,9675 0,0291 0,0907 32,7192 0,1103 7,6580 ACB 2013 0,0050 0,7761 0,0303 0,0811 32,7466 0,0740 5,5535 ACB 2014 0,0053 0,7524 0,0216 0,0741 32,8218 0,0580 6,4222 ACB 2015 0,0051 0,7738 0,0131 0,0678 32,9366 0,0871 6,9872 ACB 2016 0,0057 0,7892 0,0043 0,0640 33,0850 0,0719 6,6900 ACB 2017 0,0074 0,8224 0,0105 0,0598 33,2811 0,0778 6,9402 ACB 2018 0,0156 0,8538 0,0001 0,0682 33,4281 0,1080 7,4650 ACB 2019 0,0157 0,8720 0,0004 0,0780 33,5804 0,1231 7,3593 ACB 2020 0,0173 0,8819 0,0010 0,0867 33,7280 0,1243 2,8654 ACB 2021 0,0182 0,9526 0,0077 0,0930 33,8997 0,1699 2,5616 ACB 2022 0,0225 0,9994 0,0074 0,1064 34,0410 0,1778 8,0198 ACB 2023 0,0223 0,9990 0,0121 0,1095 34,2086 0,1952 8,0198 BID 2013 0,0074 1,1538 0,0039 0,0621 33,9380 0,1174 5,5535 BID 2014 0,0077 1,0119 0,0049 0,0539 34,1085 0,1208 6,4222 BID 2015 0,0075 1,0598 0,0082 0,0524 34,3769 0,1122 6,9872 BID 2016 0,0062 0,9968 0,0052 0,0459 34,5452 0,1050 6,6900 BID 2017 0,0058 1,0080 0,0057 0,0423 34,7230 0,1297 6,9402

BID 2018 0,0057 0,9991 0,0050 0,0473 34,8111 0,1255 7,4650 BID 2019 0,0057 1,0025 0,0060 0,0550 34,9375 0,1367 7,3593 BID 2020 0,0048 0,9899 0,0074 0,0554 34,9553 0,0970 2,8654 BID 2021 0,0062 0,9813 0,0082 0,0515 35,1051 0,1234 2,5616 BID 2022 0,0087 1,0330 0,0116 0,0517 35,2905 0,1633 8,0198 BID 2023 0,0096 1,0191 0,0126 0,0564 35,3721 0,1193 9,6366 CTG 2013 0,0101 1,0324 0,0056 0,1036 33,9878 0,1493 5,5535 CTG 2014 0,0087 1,0370 0,0112 0,0912 34,1251 0,1360 6,4222 CTG 2015 0,0073 1,0915 0,0092 0,0776 34,2897 0,1065 6,9872 CTG 2016 0,0071 1,0106 0,0105 0,0679 34,4860 0,1193 6,6900 CTG 2017 0,0068 1,0501 0,0114 0,0618 34,6296 0,1226 6,9402 CTG 2018 0,0045 1,0474 0,0001 0,0614 34,6909 0,1380 7,4650 CTG 2019 0,0076 1,0476 0,0002 0,0665 34,7545 0,1310 7,3593 CTG 2020 0,0103 1,0252 0,0010 0,0680 34,8326 0,1268 2,8654 CTG 2021 0,0093 0,9732 0,0126 0,0651 34,9651 0,1202 2,5616 CTG 2022 0,0093 1,0205 0,0124 0,0636 35,1312 0,1566 8,0198 CTG 2023 0,0099 1,0246 0,0113 1,5343 35,2481 0,1625 13,4780 EIB 2013 0,0039 1,0488 0,0049 0,0946 32,7659 0,3628 5,5535 EIB 2014 0,0003 0,8597 0,0246 0,0957 32,7130 0,2752 6,4222 EIB 2015 0,0003 0,8611 0,0019 0,1177 32,4581 0,1008 6,9872 EIB 2016 0,0024 0,8490 0,0271 0,1166 32,4893 0,1092 6,6900 EIB 2017 0,0055 0,8620 0,0227 0,1055 32,6374 0,1393 6,9402 EIB 2018 0,0043 0,8766 0,0185 0,1080 32,6592 0,1822 7,4650 EIB 2019 0,0052 0,8132 0,0171 0,1038 32,7522 0,2043 7,3593 EIB 2020 0,0067 0,7525 0,0252 0,1171 32,7089 0,2391 2,8654 EIB 2021 0,0058 0,8348 0,0196 0,1201 32,7420 0,1841 2,5616 EIB 2022 0,0159 0,8781 0,0180 0,1244 32,8517 0,1824 8,0198 EIB 2023 0,0107 0,8886 0,0265 0,0118 32,9364 0,2450 9,0198 HDB 2013 0,0025 0,7058 0,0068 0,1106 32,0880 0,1574 5,5535 HDB 2014 0,0048 0,6420 0,0063 0,0982 32,2314 0,2016 6,4222 HDB 2015 0,0059 0,7587 0,0032 0,1018 32,2990 0,1522 6,9872

HDB 2016 0,0061 0,7960 0,0025 0,0708 32,6436 0,1504 6,6900 HDB 2017 0,0103 0,8669 0,0020 0,0845 32,8745 0,1336 6,9402 HDB 2018 0,0148 0,9615 0,0014 0,0845 33,0066 0,1847 7,4650 HDB 2019 0,0175 1,1611 0,0008 0,0975 33,0668 0,1358 7,3593 HDB 2020 0,0146 1,0212 0,0016 0,0839 33,3966 0,1719 2,8654 HDB 2021 0,0172 1,1087 0,0165 0,0896 33,5569 0,1875 2,5616 HDB 2022 0,0197 1,2227 0,0167 0,1034 33,6624 0,1522 8,0198 HDB 2023 0,0172 0,9152 0,0179 0,2593 34,0318 0,2338 13,4780 KLB 2013 0,0147 0,9117 0,1293 0,0063 30,6931 0,1902 5,5535 KLB 2014 0,0076 0,8163 0,1184 0,0040 30,7710 0,1812 6,4222 KLB 2015 0,0065 0,8076 0,0113 0,0143 30,8627 0,1067 6,9872 KLB 2016 0,0040 0,8636 0,0105 0,0173 31,0471 0,1409 6,6900 KLB 2017 0,0054 0,9449 0,0195 0,0390 31,2507 0,1835 6,9402 KLB 2018 0,0055 1,0091 0,0094 0,0177 31,3760 0,1837 7,4650 KLB 2019 0,0013 1,0170 0,0122 0,0161 31,5648 0,2656 7,3593 KLB 2020 0,0022 0,8262 0,0542 0,1428 31,6790 0,2823 2,8654 KLB 2021 0,0092 0,7469 0,0189 0,0081 32,0597 0,4256 2,5616 KLB 2022 0,0063 0,8564 0,0189 0,0082 32,0826 0,3016 8,0198 KLB 2023 0,0066 0,8992 0,0193 0,0104 32,0966 0,2894 13,4780 MBB 2013 0,0127 0,6447 0,0016 0,0090 32,8261 0,1743 5,5535 MBB 2014 0,0125 0,6000 0,0273 0,0445 32,9318 0,1432 6,4222 MBB 2015 0,0114 0,6683 0,0161 0,0342 33,0294 0,1723 6,9872 MBB 2016 0,0113 0,7738 0,0132 0,1158 33,1772 0,1501 6,6900 MBB 2017 0,0111 0,8365 0,0120 0,1041 33,3800 0,1976 6,9402 MBB 2018 0,0171 0,8947 0,0133 0,1041 33,5236 0,1583 7,4650 MBB 2019 0,0196 0,9179 0,0116 0,1073 33,6508 0,1370 7,3593 MBB 2020 0,0174 0,9593 0,0109 0,1126 33,8355 0,1380 2,8654 MBB 2021 0,0218 0,9451 0,0090 0,1147 34,0398 0,1277 2,5616 MBB 2022 0,0249 1,0383 0,0109 0,1227 34,2221 0,1048 8,0198 MBB 2023 0,0223 1,0565 0,0160 0,1140 34,4822 0,1231 9,0198 MSB 2013 0,0031 0,2038 0,0012 0,0963 32,3049 0,2453 5,5535

MSB 2014 0,0014 0,1599 0,0019 0,0995 32,2790 0,1910 6,4222 MSB 2015 0,0011 0,3584 0,0043 0,1501 32,2784 0,1457 6,9872 MSB 2016 0,0015 0,2741 0,0236 0,1721 32,1594 0,1185 6,6900 MSB 2017 0,0011 0,2459 0,0223 0,1393 32,3517 0,1252 6,9402 MSB 2018 0,0063 2,0514 0,0036 0,1115 32,5566 0,1991 7,4650 MSB 2019 0,0066 0,1392 0,0027 0,1046 32,6871 0,1729 7,3593 MSB 2020 0,0114 0,1529 0,0022 0,1056 32,8055 0,1190 2,8654 MSB 2021 0,0198 0,2207 0,0174 0,1213 32,9475 0,1867 2,5616 MSB 2022 0,0217 0,0845 0,0171 0,1432 32,9913 0,2083 8,0198 MSB 2023 0,0174 1,1090 0,0287 0,1328 33,2183 0,2497 9,0198 NAB 2013 0,0047 0,0255 0,1701 0,1277 30,9908 0,2750 5,5535 NAB 2014 0,0050 0,8184 0,1311 0,0981 31,2498 0,4147 6,4222 NAB 2015 0,0055 0,8563 0,0739 0,1065 31,1997 0,2240 6,9872 NAB 2016 0,0008 0,7054 0,0430 0,0871 31,3888 0,1017 6,6900 NAB 2017 0,0044 0,9118 0,0000 0,0722 31,6281 0,1162 6,9402 NAB 2018 0,0079 0,9378 0,0001 0,0597 31,9493 0,1755 7,4650 NAB 2019 0,0077 0,9548 0,0197 0,0553 32,1816 0,1668 7,3593 NAB 2020 0,0060 0,9076 0,0083 0,0517 32,5312 0,1289 2,8654 NAB 2021 0,0094 0,8902 0,0157 0,0553 32,6630 0,1560 2,5616 NAB 2022 0,0102 0,9564 0,0163 0,0767 32,8104 0,1517 8,0198 NAB 2023 0,0125 0,9619 0,0211 0,0783 32,9776 0,1905 9,0198 NVB 2013 0,0006 0,7333 0,0115 0,1238 31,0009 0,2138 5,5535 NVB 2014 0,0002 0,6809 0,0093 0,0955 31,2375 0,2095 6,4222 NVB 2015 0,0001 0,6004 0,0122 0,0715 31,5070 0,1767 6,9872 NVB 2016 0,0002 0,6066 0,0459 0,0491 31,8653 0,1946 6,6900 NVB 2017 0,0003 0,7023 0,0129 0,0469 31,9055 0,1680 6,9402 NVB 2018 0,0005 0,7566 0,0101 0,0467 31,9135 0,1182 7,4650 NVB 2019 0,0005 0,6415 0,0114 0,0566 32,0180 0,1990 7,3593 NVB 2020 0,0000 0,5592 0,0171 0,0500 32,1264 0,1558 2,8654 NVB 2021 0,0000 0,6450 0,0300 0,0613 31,9321 0,0733 2,5616 NVB 2022 0,0000 0,6688 0,1793 0,0686 32,1291 0,1758 8,0198

NVB 2023 -0,0070 0,7061 0,2976 0,0559 32,1980 0,1142 13,4780 OCB 2013 0,0074 1,0556 0,0653 0,1375 31,1213 0,1314 5,5535 OCB 2014 0,0056 0,8981 0,0126 0,1145 31,2970 0,1008 6,4222 OCB 2015 0,0042 0,9386 0,0015 0,0934 31,5319 0,1509 6,9872 OCB 2016 0,0061 0,8942 0,0175 0,0798 31,7870 0,0922 6,6900 OCB 2017 0,0097 0,9056 0,0179 0,0785 32,0654 0,1677 6,9402 OCB 2018 0,0176 0,9330 0,0229 0,1126 32,2358 0,1541 7,4650 OCB 2019 0,0219 1,0282 0,0184 0,1262 32,4031 0,1716 7,3593 OCB 2020 0,0232 1,0237 0,0169 0,1291 32,6584 0,1403 2,8654 OCB 2021 0,0239 1,0329 0,0132 0,1340 32,8486 0,1472 2,5616 OCB 2022 0,0181 1,1722 0,0223 0,1498 32,8988 0,1271 8,0198 OCB 2023 0,0138 1,1489 0,0265 0,1349 33,1121 0,1734 13,4780 SSB 2013 0,0019 0,5784 0,1785 0,0772 32,0114 0,4148 5,5535 SSB 2014 0,0011 0,7121 0,0018 0,0763 32,0153 0,3748 6,4222 SSB 2015 0,0011 0,7507 0,0024 0,0730 32,0708 0,2188 6,9872 SSB 2016 0,0011 0,8178 0,0014 0,0603 32,2693 0,1769 6,6900 SSB 2017 0,0024 0,8811 0,0030 0,0520 32,4594 0,1710 6,9402 SSB 2018 0,0035 0,9948 0,0046 0,0628 32,5761 0,1556 7,4650 SSB 2019 0,0070 1,0302 0,0045 0,0746 32,6898 0,1901 7,3593 SSB 2020 0,0075 0,9611 0,0043 0,0821 32,8251 0,1672 2,8654 SSB 2021 0,0123 1,1622 0,0165 0,0967 32,9860 0,2290 2,5616 SSB 2022 0,0175 1,3324 0,0225 0,1278 33,0753 0,2417 8,0198 SSB 2023 0,0010 1,2205 0,0194 0,1285 33,2150 0,0309 9,0198 SGB 2013 0,0118 0,9877 0,0006 0,3130 30,3178 0,0738 5,5535 SGB 2014 0,0114 0,9484 0,0035 0,2825 30,3925 0,0558 6,4222 SGB 2015 0,0024 0,8836 0,0903 0,2362 30,5073 0,1149 6,9872 SGB 2016 0,0073 0,8846 0,0966 0,2263 30,5780 0,1343 6,6900 SGB 2017 0,0257 0,2765 0,4369 0,1909 30,6906 0,1924 6,9402 SGB 2018 0,0020 0,9314 0,1346 0,2028 30,6453 0,1657 7,4650 SGB 2019 0,0063 0,9291 0,1763 0,1850 30,7583 0,2598 7,3593 SGB 2020 0,0041 0,8477 0,1239 0,1782 30,8067 0,2703 2,8654

SGB 2021 0,0050 0,9114 0,0197 0,1775 30,8341 0,2580 2,5616 SGB 2022 0,0069 0,9129 0,0212 0,1638 30,9524 0,2234 8,0198 SGB 2023 0,2325 0,8400 0,0203 0,1483 31,0810 3,7994 13,4780 SHB 2013 0,0059 0,8430 0,0673 0,0777 32,5982 0,2283 5,5535 SHB 2014 0,0047 0,8447 0,0409 0,0661 32,7611 0,1990 6,4222 SHB 2015 0,0039 0,8831 0,0441 0,0582 32,9526 0,1762 6,9872 SHB 2016 0,0039 0,9748 0,0460 0,0599 33,0861 0,1460 6,6900 SHB 2017 0,0054 1,0174 0,0377 0,0541 33,2870 0,1353 6,9402 SHB 2018 0,0052 0,9634 0,0319 0,0532 33,4095 0,1099 7,4650 SHB 2019 0,0066 1,0229 0,0234 0,0534 33,5316 0,1249 7,3593 SHB 2020 0,0063 1,0068 0,0183 0,0618 33,6537 0,1155 2,8654 SHB 2021 0,0099 1,1076 0,0169 0,0754 33,8588 0,1586 2,5616 SHB 2022 0,0140 1,0662 0,0281 0,0845 33,9426 0,1453 8,0198 SHB 2023 0,0058 0,9595 0,0302 0,0863 34,0775 0,0830 9,0198 STB 2013 0,0138 0,8399 0,0090 0,1182 32,7148 0,0929 5,5535 STB 2014 0,0116 0,7851 0,0075 0,1052 32,8770 0,0671 6,4222 STB 2015 0,0022 0,7123 0,0083 0,0818 33,3079 0,0558 6,9872 STB 2016 0,0003 0,6818 0,0100 0,0716 33,4362 0,0521 6,6900 STB 2017 0,0032 0,6970 0,0467 0,0673 33,5404 0,0450 6,9402 STB 2018 0,0044 0,7345 0,0213 0,0646 33,6375 0,0513 7,4650 STB 2019 0,0054 0,7385 0,0004 0,0627 33,7482 0,0739 7,3593 STB 2020 0,0054 0,7951 0,0004 0,0625 33,8305 0,0734 2,8654 STB 2021 0,0065 0,9077 0,0150 0,0704 33,8870 0,0552 2,5616 STB 2022 0,0085 0,9646 0,0098 0,0698 34,0144 0,0768 8,0198 STB 2023 0,0114 0,9303 0,0228 0,0727 34,1448 0,1049 9,0198 TCB 2013 0,0041 0,5857 0,0075 0,0960 32,6993 0,1293 5,5535 TCB 2014 0,0062 0,6098 0,0238 0,0931 32,8009 0,1297 6,4222 TCB 2015 0,0080 0,7887 0,0166 0,0938 32,8885 0,1052 6,9872 TCB 2016 0,0134 0,8222 0,0158 0,0908 33,0922 0,1151 6,6900 TCB 2017 0,0239 0,9408 0,0161 0,1111 33,2272 0,1365 6,9402 TCB 2018 0,0264 0,7941 0,0175 0,1924 33,4024 0,1518 7,4650

TCB 2019 0,0267 0,9979 0,0001 0,1930 33,5809 0,1460 7,3593 TCB 2020 0,0286 1,0002 0,0007 0,2044 33,7169 0,0976 2,8654 TCB 2021 0,0324 1,1035 0,0066 0,1956 33,9744 0,1389 2,5616 TCB 2022 0,0292 1,1733 0,0072 0,1937 34,1807 0,1410 8,0198 TCB 2023 0,0214 1,1272 0,0116 0,1833 34,3756 0,1587 9,0198 VCB 2013 0,0093 0,8256 0,0273 0,0994 33,7816 0,2615 5,5535 VCB 2014 0,0080 0,7658 0,0231 0,0813 33,9888 0,2906 6,4222 VCB 2015 0,0079 0,7736 0,0184 0,0718 34,1448 0,2369 6,9872 VCB 2016 0,0087 0,7804 0,0151 0,0650 34,3004 0,2271 6,6900 VCB 2017 0,0088 0,7670 0,0114 0,0535 34,5735 0,3252 6,9402 VCB 2018 0,0136 0,7879 0,0016 0,0615 34,6102 0,2550 7,4650 VCB 2019 0,0152 0,7913 0,0020 0,0709 34,7399 0,2436 7,3593 VCB 2020 0,0139 0,8137 0,0024 0,0764 34,8211 0,2384 2,8654 VCB 2021 0,0156 0,8462 0,0064 0,0836 34,8859 0,1882 2,5616 VCB 2022 0,0165 0,9209 0,0068 0,0808 35,1342 0,2341 8,0198 VCB 2023 0,0180 0,8896 0,0098 0,0985 35,1483 0,2224 13,4780 VIB 2013 0,0007 0,8150 0,0011 0,1159 31,9732 0,1249 5,5535 VIB 2014 0,0065 0,7783 0,0030 0,1178 32,0213 0,1209 6,4222 VIB 2015 0,0062 0,8963 0,0080 0,1138 32,0655 0,1009 6,9872 VIB 2016 0,0054 1,0155 0,0258 0,0913 32,2804 0,1342 6,6900 VIB 2017 0,0091 1,1680 0,0249 0,0768 32,4445 0,1208 6,9402 VIB 2018 0,0158 1,1329 0,0251 0,0830 32,5667 0,0861 7,4650 VIB 2019 0,0177 1,0559 0,0256 0,0785 32,8488 0,1312 7,3593 VIB 2020 0,0190 1,1275 0,0208 0,0793 33,1310 0,1242 2,8654 VIB 2021 0,0207 1,1610 0,0232 0,0852 33,3660 0,1757 2,5616 VIB 2022 0,0247 1,1590 0,0245 0,1053 33,4682 0,1855 8,0198 VIB 2023 0,0209 1,1078 0,0314 0,1020 33,6469 0,1905 9,0198 VPB 2013 0,0084 0,6259 0,0281 0,0681 32,4290 0,1248 5,5535 VPB 2014 0,0077 0,7234 0,0254 0,0582 32,7263 0,1163 6,4222 VPB 2015 0,0124 0,8966 0,0269 0,0742 32,8982 0,0954 6,9872 VPB 2016 0,0172 1,1687 0,0291 0,0812 33,0637 0,0616 6,6900

VPB 2017 0,0232 1,3678 0,0339 0,1197 33,2578 0,0956 6,9402 VPB 2018 0,0228 1,2992 0,0350 0,1204 33,4096 0,0905 7,4650 VPB 2019 0,0219 1,2021 0,0342 0,1260 33,5638 0,0690 7,3593 VPB 2020 0,0249 1,2458 0,0341 0,1442 33,6690 0,0683 2,8654 VPB 2021 0,0210 1,4691 0,0457 0,1871 33,9362 0,1284 2,5616 VPB 2022 0,0268 1,4459 0,0573 0,1962 34,0783 0,0960 8,0198 VPB 2023 0,0104 1,2466 0,0502 0,2063 34,3374 0,1282 9,0198

APPENDIX 3 Descriptive statistics of variables

Variable Obs Mean Std dev Min Max

sum ROA LLP LTD NPL LEV SIZE GDP

ROA LLP LTD NPL LEV SIZE GDP

corr ROA LLP LTD NPL LEV SIZE GDP

Pooled Ordinary Least Squares Model (Pooled OLS)

_cons -.0645425 0219886 -2.94 0.004 -.1078857 -.0211994 GDP -.0015938 0038351 -0.42 0.678 -.0091535 0059659 SIZE 0019735 0006729 2.93 0.004 0006472 0032998 LEV -.0000638 0000277 -2.31 0.022 -.0001183 -9.33e-06 NPL -.0298593 0161687 -1.85 0.066 -.0617306 0020119 LTD 0088237 0029704 2.97 0.003 0029687 0146788 LLP 3774202 0193802 19.47 0.000 3392187 4156217 ROA Coefficient Std err t P>|t| [95% conf interval]

Total 061061344 219 000278819 Root MSE = 00999 Adj R-squared = 0.6421 Residual 021254431 213 000099786 R-squared = 0.6519 Model 039806913 6 006634485 Prob > F = 0.0000 F(6, 213) = 66.49 Source SS df MS Number of obs = 220 reg ROA LLP LTD NPL LEV SIZE GDP

F test that all u_i=0: F(19, 194) = 8.01 Prob > F = 0.0000 rho 56971145 (fraction of variance due to u_i) sigma_e 00783482 sigma_u 00901522

_cons -.2201706 0406793 -5.41 0.000 -.3004011 -.1399401 GDP -.0008921 0030315 -0.29 0.769 -.0068709 0050868 SIZE 0067702 0012635 5.36 0.000 0042783 0092621 LEV 0000626 0000315 1.99 0.048 4.38e-07 0001248 NPL -.0530647 0139736 -3.80 0.000 -.0806244 -.025505 LTD 0037678 0031956 1.18 0.240 -.0025348 0100705 LLP 3851452 0160373 24.02 0.000 3535154 416775 ROA Coefficient Std err t P>|t| [95% conf interval] corr(u_i, Xb) = -0.2937 Prob > F = 0.0000 F(6,194) = 109.06

Overall = 0.5508 max = 11 Between = 0.0029 avg = 11.0 Within = 0.7713 min = 11 R-squared: Obs per group:

Group variable: BANK Number of groups = 20Fixed-effects (within) regression Number of obs = 220 xtreg ROA LLP LTD NPL LEV SIZE GDP, fe

est sto rem rho 36711365 (fraction of variance due to u_i) sigma_e 00783482 sigma_u 00596714

_cons -.1429899 0315582 -4.53 0.000 -.2048428 -.081137 GDP -.0014328 0030988 -0.46 0.644 -.0075064 0046409 SIZE 0043786 0009759 4.49 0.000 0024659 0062913 LEV 0000308 0000301 1.02 0.307 -.0000282 0000898 NPL -.0446822 0140095 -3.19 0.001 -.0721403 -.0172241 LTD 0063355 0030208 2.10 0.036 0004148 0122562 LLP 3870833 0162365 23.84 0.000 3552605 4189062 ROA Coefficient Std err z P>|z| [95% conf interval] corr(u_i, X) = 0 (assumed) Prob > chi2 = 0.0000 Wald chi2(6) = 612.60

Overall = 0.6155 max = 11 Between = 0.0348 avg = 11.0 Within = 0.7661 min = 11 R-squared: Obs per group:

Group variable: BANK Number of groups = 20Random-effects GLS regression Number of obs = 220 xtreg ROA LLP LTD NPL LEV SIZE GDP, re

LM-Breusch and Pagan Multiplier test results

Test of H0: Difference in coefficients not systematic

B = Inconsistent under Ha, efficient under H0; obtained from xtreg. b = Consistent under H0 and Ha; obtained from xtreg. GDP -.0008921 -.0014328 0005407

LLP 3851452 3870833 -.0019381 fem rem Difference Std err.

Prob > F = 0.1653 F( 1, 19) = 2.083 H0: no first-order autocorrelation Wooldridge test for autocorrelation in panel data xtserial ROA LLP LTD NPL LEV SIZE GDP

Prob>chi2 = 0.0000 chi2 (20) = 1125.20 H0: sigma(i)^2 = sigma^2 for all i in fixed effect regression modelModified Wald test for groupwise heteroskedasticity xttest3