Luận văn factors affecting the net interest margin of vietnamese commercial banks

MINISTRY OF EDUCATION AND TRAINING THE STATE BANK OF VIETNAM HO CHI MINH UNIVERSITY OF BANKING DANG NGUYEN PHUONG LINH FACTORS AFFECTING THE NET INTEREST MARGIN OF VIETNAMESE COMMERCI

INTRODUCTION

Background

Commercial banks play an essential role in the economy, acting as a capital circulation system and driving economic growth of a country (Puspitasari et al., 2021)

Against the backdrop of rapid international integration, Vietnam faces significant opportunities for broad economic growth and, in particular, the expansion of its finance and banking sector Vietnamese commercial banks, however, confront intense competition for market share and technology leadership from both domestic peers and foreign financial entities This challenge is intensified by rising global inflation driven by spikes in goods and raw materials due to geopolitical conflicts and disrupted supply chains In this context, the banking sector plays a key role in economic integration by acting as financial intermediaries between savers and borrowers, with any inefficiency risking idle capital, weaker bank performance, and broader economic health Amid these pressures, ensuring stability and sustainable development requires thorough assessment of bank performance, a priority for regulators and government agencies alike.

Net Interest Margin (NIM) is a core indicator of banking performance, capturing how efficiently a financial institution earns income from its interest-earning assets relative to its interest-bearing liabilities It represents the interest rate spread generated by a bank’s core activities—deposit mobilization, lending, and investments—that determine profitability Net interest income (NII) is central to measuring interest rate trends and income variation across banks Because NIM directly ties to a bank’s fundamental operations, it helps institutions monitor the profitability of their assets and identify the lowest-cost sources of funding.

Profit maximization is commonly the main objective for commercial banks, but chasing high profits can compromise their role as financial intermediaries Low deposit rates may limit deposit mobilization, while high lending rates can dampen investment opportunities and erode banks’ competitiveness in an era of ongoing economic integration Banks with high net interest margins often impose higher borrowing costs, potentially discouraging private investment and slowing economic development Therefore, maintaining a reasonable NIM is an essential requirement for central banks and commercial banks alike Efficient banking supports the smooth circulation of capital, generates resources for the state, creates employment, and strengthens connections among economic agents, making it imperative to balance NIM with competitiveness to ensure the effective operation of banks and broader access to credit for businesses and households, thereby driving sustainable long-term national economic growth.

Research findings provide a scientific basis for managers of Vietnamese commercial banks to make informed decisions and design policies that boost operational efficiency and contribute to overall economic development This study, titled “Factors Influencing the Net Interest Margin of Vietnamese Commercial Banks,” analyzes the key determinants of the net interest margin (NIM) in Vietnam’s banking sector, offering actionable insights for improving profitability, risk management, and financial stability.

Research objectives

This study investigates the factors influencing the net interest margin (NIM) of Vietnamese commercial banks from 2012 to 2023, clarifying how funding costs, asset mix, and macroeconomic conditions shape NIM Drawing on the findings, it proposes managerial implications to boost NIM for Vietnamese banks, including strategies in asset-liability management, pricing, and risk controls to sustain profitability.

Followed are a few specific objectives that has been set out in order to achieve the research goals:

(i) Identifying and clarify the factors affecting the NIM of Vietnamese Commercial Banks

(ii) Measuring the extent to which these factors influence the Net Interest Margin of Vietnamese Commercial Banks

(ii) Proposing managerial implications aimed at improving the Net Interest Margin of Vietnamese Commercial Banks.

Research questions

This thesis is aimed to solve these research questions:

(i) What factors affect the Net Interest Margin of Vietnamese Commercial Banks?

(ii) To what extent do these factors affect the Net Interest Margin of Vietnamese Commercial Banks?

(iii) What are the appropriate managerial implications to improve the Net Interest Margin of Vietnamese Commercial Banks?

Subject and scope of the research

Research subject: The NIM ratio and the factors influencing the NIM of commercial banks in Vietnam

Research scope: The study covers 28 commercial banks in Vietnam over a 12-year period, from 2012 to 2023

As of January 28, 2024, Vietnam had a total of 49 licensed banks, including commercial banks, joint venture banks, and banks with 100% foreign ownership This study, however, limits its scope to 28 Vietnamese Commercial Banks and does not cover the entire banking system The restriction arises from data collection challenges, notably accessing financial statements of joint venture banks and wholly foreign‑owned banks, as well as mismatches in their financial reporting structures compared with Vietnamese Commercial Banks Additionally, several Vietnamese Commercial Banks do not have complete financial statements for 2012–2023 due to later establishment or mergers with other banks.

From 2012 to 2023, the Vietnamese economy experienced pivotal milestones and structural reforms that shaped its post-crisis economic development 2012 was a turning point under the Party’s 11th Congress, emphasizing growth-model innovation, strengthening the financial system, restructuring state-owned enterprises, and improving macroeconomic efficiency The period from 2020 to 2023 brought substantial challenges due to the COVID-19 pandemic, which disrupted global supply chains and slowed economic growth, yet Vietnam sustained positive growth rates and gradually recovered through adaptive measures.

Research methodology

This study relies on secondary data collected from the financial statements and annual reports of the State Bank of Vietnam and Vietnamese commercial banks published over the years These data are highly reliable, having been audited by reputable firms including KPMG, Ernst & Young, Deloitte, and PwC Macroeconomic indicators such as growth and inflation rates were also sourced from the official websites of the World Bank and the International Monetary Fund (IMF).

Qualitative method: This study compares the results of empirical analysis with findings from previous related research in order to explain the research objectives, address the research questions

Using a quantitative approach rooted in multivariate regression, the author combines several estimators: pooled OLS to estimate the baseline model, followed by fixed effects (FEM) and random effects (REM) models F-tests and Hausman tests are employed to identify the most appropriate specification among OLS, FEM, and REM To overcome potential model deficiencies, Feasible Generalized Least Squares (FGLS) is applied, and the Generalized Method of Moments (GMM) is used to address endogeneity that conventional estimators cannot resolve Data are processed with Stata version 17.0 to estimate the regression coefficients, and the resulting results are used to test the research hypotheses and identify the impact of various factors on the Net Interest Margin (NIM) of Vietnamese commercial banks.

Contribution of the research

From a scientific perspective: This study clarifies and quantifies the extent to which various factors influence the Net Interest Margin (NIM) of Vietnamese commercial banks during the period from 2012 to 2023

From a practical perspective: The author has collected and utilized the most up- to-date data as of 2023 regarding the determinants of NIM over a 12-year period from

Spanning 2012 to 2023, the study improves the empirical accuracy and timeliness of the data used in the thesis compared with earlier research The updated empirical findings provide bank managers with a more comprehensive understanding of the factors influencing net interest margin (NIM), enabling them to derive practical managerial implications to improve NIM and enhance operational efficiency in commercial banks.

Research structure

This chapter establishes the research background and rationale, clearly states the study’s research questions and objectives, defines the research object and its scope, describes the data sources and methodological approach, highlights the study’s contributions to the field, and outlines the overall structure of the thesis.

THEORETICAL BACKGROUND AND EMPIRICAL STUDIES

Theoretical Basis for Net Interest Margin

Net interest margin (NIM) is a widely used metric that measures changes and trends in interest rate spreads, defined as the percentage difference between a bank’s total interest income and total interest expenses It is typically expressed as a percentage over a specific time frame, usually a quarter or a year, which helps compare performance across periods In the banking sector, NIM captures the interest spread earned from core operations—mobilizing funds, investing, and lending—and thus serves as a key indicator of profitability and growth potential While several methods exist to calculate NIM, two formulas are most commonly used by researchers and bank managers, enabling consistent benchmarking Rose (2001) describes NIM as a useful tool for monitoring shifts in rate spreads, Fungáčová & Poghosyan (2011) define it as the percentage difference between interest income and interest expenses, and Golin (2001) emphasizes its expression as a percentage over a defined time frame, highlighting its practical applicability for quarterly or annual analysis.

NIM = Interest Income−Interest Expense

Total Earning Asset = Net Interest Income

NIM = Interest Income−Interest Expense

Total Assets = Net Interest Income

In this thesis, the author adopts formula (1), using the percentage of Net Interest Income over Earning Assets to measure the NIM of Vietnamese commercial banks

Net interest income (NII) defines the earnings a bank derives from the spread between its deposit (cost of funds) and lending (interest earned) rates It is calculated as Interest Income and Similar Income minus Interest Expense and Similar Charges, as reflected in the bank's income statement This measure captures how effectively a financial institution leverages the gap between what it pays on deposits and what it earns on loans, and it is a key indicator of a bank's core profitability and interest-rate sensitivity.

• Interest Income includes earnings from lending to customers and financial entities, deposits, debt securities, guarantees, and other interest-generating activities

• Interest Expenses include payments on deposits, borrowings, issuance of valuable papers, and other credit-related costs

• Earning Assets include items such as deposits at the State Bank of Vietnam (SBV), interbank loans and deposits (excluding provisioned risks), trading securities (net of provisions), customer loans (net), and investment securities (net of markdowns), as defined in Vietnam’s financial reporting standards for credit institutions

“Among the various economic indicators used to evaluate the profitability of banks, the Net Interest Margin (NIM) is one of the most widely applied metrics It serves as a key measure to determine whether the core business activities of commercial banks namely, deposit mobilization and lending are being managed efficiently According to Goblin (2001): "NIM assesses the balance between interest income and interest expense" providing insight into the effectiveness of banks' interest-based operations

The net interest margin (NIM) ratio reflects a bank’s capacity, guided by its board of directors and staff, to sustain revenue growth from loans and investments in relation to cost increases—primarily the interest paid on deposits and money market borrowings By analyzing NIM, commercial banks can optimize their interest-earning assets and identify low-cost funding sources, enabling efficient financial intermediation at the lowest possible cost Goblin (2001) cautions that evaluating whether a NIM ratio is good or bad must be done carefully, taking into account both microeconomic and macroeconomic factors.

From a profitability perspective, a high net interest margin (NIM) signals solid earnings, reflecting substantial interest income relative to earning assets It suggests the bank is adept at deploying financial instruments and managing its operations efficiently Moreover, a higher NIM can bolster profitability across the commercial banking system, strengthen capitalization, and fortify the bank’s financial position by creating a buffer against economic shocks (Barajas, 2000).

A low net interest margin (NIM) signals that a bank is struggling to efficiently balance its interest-earning assets This can occur when funding costs exceed the income generated from loans and other assets, narrowing the margin between what the bank earns and pays It can also stem from an over-reliance on non-interest income sources that are volatile and not central to the core functions of a commercial bank.

Profit maximization remains the primary objective for commercial banks, but from a broader societal perspective, a consistently high net interest margin (NIM) over time may reflect elevated lending rates that hinder access to credit; this tension is discussed by Zuzana and Tigran (2008).

A persistently high net interest margin (NIM) undermines banks’ financial intermediation, reducing investment opportunities and weakening the banking system’s competitiveness Conversely, offering deposits at low interest rates may discourage saving and impede economic growth (Almarzoqi, 2015).

Determining whether a high or low net interest margin (NIM) is favorable remains a nuanced issue that requires evaluation across multiple dimensions (Doliente, 2005) A high NIM can signal effective management of earning assets and bolster bank profitability, but if sustained, it may impede credit access for economic stakeholders At times, a high NIM results not from higher interest income but from cost reduction and efficient asset management Moreover, NIM is influenced by banks' operational strategies, macroeconomic conditions, and broader economic challenges.

Commercial banks should strive to maintain an optimal net interest margin (NIM) that balances profitability with broader social objectives By achieving this balance, they can strengthen competitiveness, fulfill their essential role as financial intermediaries, and contribute to sustainable, long-term economic development.

Foundational Theories Related to NIM

Theory of Economies of Scope

Expanding a firm’s product and service portfolio tends to lower average input costs, as resources can be reallocated and used more efficiently while common input factors are shared across multiple business segments Diversification enhances operational efficiency by flexibly leveraging these shared resources, enabling cost savings and more agile execution When different activities draw on a similar set of inputs, total costs decrease, contributing to improved overall operational performance.

A 2016 study in Vietnam by Tran Chi Chinh and Nguyen Huu Tien shows that an extensive branch network is a key driver of economies of scale for commercial banks, enabling them to reduce costs and improve operational efficiency.

Panzar and Willig's 1977 theory of economies of scope explains why banks are expanding beyond traditional services like deposit-taking and lending to offer a broader mix of non-credit services, including insurance, asset management, investment consulting, and international payments By leveraging existing resources such as branch networks and information technology systems, banks can improve efficiency and cross-sell opportunities while reducing reliance on a single income source, notably interest income tied to lending and economic cycles In today’s volatile financial markets and under stricter risk-management standards, this diversification strengthens resilience and creates new revenue streams for banks.

The business cycle, or economic cycle, describes how the economy responds to real shocks—such as technological change, natural disasters, and wars—with positive or negative fluctuations that ripple through production, consumption, and investment These cycles are seen as recurring changes in real GDP, unfolding through four main phases: recession, crisis, recovery, and boom (prosperity) Each phase reflects shifts in output, demand, and investment over time, shaping the economy’s overall trajectory.

During the expansion phase of the financial and banking sector, productivity gains and rising market confidence—often triggered by a technological breakthrough or policy easing—spur stronger credit activity As loan demand increases, interest rates remain stable, non-performing loans stay low, and banks’ net interest margins (NIM) improve.

When adverse shocks—such as war, financial crises, or pandemics—hit an economy, investment and production contract, leading to a drop in credit demand and higher credit risk, which drives up non-performing loans Under these conditions, banks' net interest margins (NIM) tend to shrink, profitability declines, and maintaining adequate capital and liquidity becomes a major challenge.

Banks, as financial intermediaries, both transmit and absorb macroeconomic shocks, and their ability to respond to macroeconomic risks while adjusting lending policies is a critical driver of economic stability during recessions and of recovery in the subsequent phases of the business cycle.

Factors Influencing Net Interest Margin

To help commercial banks maximize both financial performance and social value, it is essential to identify the determinants of the Net Interest Margin (NIM) This study categorizes these determinants into two overarching groups: microeconomic factors, including bank‑specific characteristics and lending and funding dynamics, and macroeconomic factors, such as interest rate movements, inflation, and overall economic conditions Understanding these microeconomic and macroeconomic drivers enables banks to fine‑tune their strategies to boost profitability while delivering broader social benefits.

2.3.1 Factors of the banking industry

Bank size is a key indicator of the differences in how banks of various scales operate in the market Large banks benefit from extensive operational networks and a strong industry reputation, enabling them to attract a broader base of potential customers They also offer a wider range of financial products and services, and this diversification supports lower-cost fund mobilization and higher returns compared with smaller banks.

SIZE (Bank Size) = ln (Total Assets)

Equity Capital (CAP) is a key indicator of a commercial bank’s financial strength and risk aversion, reflecting the level of own funds available to back lending and other activities As Mishkin (2010) notes, "Equity capital consists of the initial capital contributions from shareholders and the retained earnings accumulated during the bank’s operations." A higher level of equity capital means more of the bank’s own capital supports its business, reducing reliance on externally mobilized funds and lowering capital-raising costs This, in turn, can reduce interest expenses and related costs, potentially increasing the bank’s net interest margin (NIM).

Moreover, strong equity capital levels enhance a bank’s resilience against potential business risks, serving as a buffer to guard against insolvency and financial distress

Extensive evidence links equity capital to net interest margin (NIM), with key findings reported by Maudos & Guevara (2004), Saunders & Schumacher (2000), Doliente (2005), Maudos & Solís (2009), Ugur & Erkus (2010), Kasman (2010), Fungáčová & Poghosyan (2011), and Hawtrey & Liang (2008) In line with this literature, the present study treats the equity capital ratio as a primary measure of capital strength, defined as the proportion of equity capital to total assets.

CAP (Equility Capital) = Equity Capital

Commercial banks function as financial intermediaries, channeling idle capital from savers to borrowers in the market, with lending as a core and highly profitable activity The interest earned from lending, after deducting interest paid on mobilized funds and other related operating costs, constitutes a major source of bank profit As Hamadi & Awdeh (2012) state, when the loan size expands, the bank’s interest income tends to increase accordingly, which in turn raises the Net Interest Margin (NIM).

Lending activities inherently carry credit risks, particularly through non-performing loans (NPLs), which can adversely affect bank performance and compress net interest margins (NIM) To capture the scale of lending activity, this study measures loan size as the proportion of total loans to total assets, a metric that reflects how much of a bank’s assets are funded by loans and helps assess exposure to credit risk and its impact on profitability.

LOAN (Loan Size) = Total Loans

Liquidity (LIQ) in finance is the ability to convert assets into cash quickly and at minimal cost In the banking sector, liquidity refers to a bank’s capacity to meet its financial obligations as they come due, such as honoring withdrawal requests or disbursing pre-approved credit lines As Alshatti (2015) notes: 'In banking, liquidity is the ability to meet financial obligations as soon as they come due, including cash withdrawals and disbursement of pre-approved loans'.

Evaluating a bank’s liquidity means assessing its ability to meet its financial obligations promptly and without delay Although liquidity risk is a central concern for commercial banks, it can pose a serious threat in certain circumstances As Đặng Văn Dân (2015) notes, liquidity risk arises when a bank cannot meet its financial obligations on time or must raise funds at high costs to do so To attract more deposits and bolster liquidity, banks may raise interest rates, which increases funding costs and can reduce net interest margin (NIM).

Studies by Doliente (2005), Hamadi (2012), and Bektas (2014) have shown that liquidity significantly affects the net interest margin (NIM) In line with Bektas (2014), this study adopts the liquidity ratio used in prior research to quantify bank liquidity, enabling a rigorous analysis of how liquidity levels shape NIM dynamics.

LIQ (Liquidity) = Cash and Cash Equivalents

State bank’s reserve policy (SBRP): According to Clause 1, Article 14 of the

Under the Law on the State Bank of Vietnam (2010), required reserves are the funds that credit institutions must deposit with the State Bank to implement national monetary policy As commercial banks mobilize deposits from customers, they must not only reserve an adequate amount of cash for liquidity but also allocate a portion of these deposits to a reserve account at the State Bank of Vietnam (SBV) The average balance in this reserve account must not fall below the reserve requirement ratio established by the SBV.

As Bektas (2014) states, "An increase in required reserves means less money available for the bank’s operational activities, which consequently reduces the Net Interest Margin (NIM)" Saunders & Schumacher (2000) counter with, "Higher reserve requirements may drive banks to demand a higher NIM to compensate for the reduced profitability due to funds being tied up in non-earning reserves".

Numerous studies, including those by Pham Minh Dien (2018) and Saunders & Schumacher (2000), have explored the relationship between required reserves and NIM

In this study, the required reserve variable is measured as the ratio of cash and deposits at the SBV to total assets:”

SBRP (State bank’s reserve policy) = "Cash and Deposits at the SBV"

Credit Risk (LLR) sits at the heart of a globally integrated, marketized financial system where credit activities and banking services are deeply interwoven In this environment, a healthy credit market is essential for safeguarding the safety of credit institutions and for maintaining overall economic stability Although lending operations generate the majority of income for commercial banks, they also introduce significant risks that can undermine banking efficiency and profitability.

Credit risk is the possibility that a borrower will fail to fulfill their financial obligations when due Waweru and Kalami (2009) note that non-performing loans are closely linked to banking sector crises As banks expand lending activities, they face higher credit risk, prompting higher loan loss provisions Consequently, interest-related expenses rise and can erode the Net Interest Margin (NIM).

Numerous empirical studies have demonstrated a relationship between credit risk and NIM, including those by Angbazo (1997), Maudos & Guevara (2004), and Tarus

(2012) According to Tarus (2012), "Credit risk is measured by the ratio of loan loss provisions to total loans of a bank"

LLR (Loan Loss Reserves) = "Loan Loss Provisions"

State-owned commercial banks typically operate with greater stability and a lower risk appetite due to stringent regulatory oversight Their primary focus is capital preservation and risk mitigation, favoring safer lending practices such as consumer loans and secured loans over high-risk investment projects While this conservative approach strengthens financial security, it can curb interest income growth because low-risk loans generally carry lower interest rates, thereby limiting overall bank profitability.

A 2022 study by Nguyen Thanh Binh reveals a negative relationship between state ownership (SOC) and net interest margin (NIM) The findings suggest that banks with higher levels of state ownership tend to have lower NIMs, a pattern attributed to more cautious lending practices associated with state-controlled banks.

SOC (State Ownership Characteristics): Agribank, BIDV, Vietcombank, Vietinbank (more than 51% state ownership) are assigned a value of 1, while other commercial banks are assigned a value of 0

Economic Growth Rate (GDP): GDP also known as Gross Domestic Product, is a macroeconomic indicator that measures the overall economic activity of a country

Literature Review

Ho and Saunders (1981) introduced a seminal, foundational model of net interest margin (NIM) for commercial banks, which has shaped the theoretical framework for many later studies The model shows that NIM is influenced by the bank’s level of risk aversion, the volume of its transactions, the structure of the banking market, and the variance of interest rates It emphasizes that the analysis must consider the structure of both assets and liabilities together, since they are interconnected through transaction uncertainty.

Building upon the model of Ho & Saunders (1981), Saunders & Schumacher

(2000) applied panel data regression techniques to investigate the determinants of NIM for banks across various European and American countries Their study covered a sample of 110 banks in the United States, 32 in the United Kingdom, 94 in Switzerland,

135 in Italy, 110 in France, 151 in Germany, and 114 in Spain during the period from

1988 to 1995 The independent variables in their model included: risk aversion, operating costs, credit risk, implicit interest payments, market structure, and reserve requirements mandated by central banks The empirical findings revealed that reserve requirements, implicit interest payments, risk aversion, credit risk, market structure, and interest rate volatility all had a positive relationship with the NIM Among these, interest rate volatility was identified as the most decisive factor influencing the NIM

In a 2012 study, Hamadi and Awdeh analyze how banking sector characteristics and macroeconomic variables affect the net interest margin (NIM) of 53 Lebanese banks—32 domestic and 21 international—over 1996–2009 They find pronounced NIM differences between domestic and foreign banks Among domestic banks, NIM falls with larger bank size, lower management quality, and greater liquidity; capitalization and credit risk also reduce NIM, though to a smaller extent Risk aversion, ownership type, market structure, and the use of dollar-denominated loans and deposits are inversely related to NIM By contrast, GDP growth, deposit growth, higher operating costs, larger loan sizes, higher central bank discount rates, inflation, national savings, and domestic investment tend to raise NIM, with interbank lending rates having a smaller positive effect For foreign banks, size, liquidity, capitalization, and credit risk do not significantly influence NIM The study also shows that domestic markets, macroeconomic conditions, banking-sector characteristics, central bank discount rates, and interbank rates have a weaker impact on NIM for foreign banks than for domestic ones.

Ong (2013) investigated how bank-specific characteristics and macroeconomic variables influence the financial performance of Malaysian commercial banks over 2003–2009 Using profitability indicators—Net Interest Margin (NIM), Return on Assets (ROA), and Return on Equity (ROE)—and regression analyses to identify determinants, the study found that risk aversion, bank size, and liquidity are positively associated with NIM, while loan loss provisions and the cost-to-income ratio have negative effects on NIM GDP and inflation, by contrast, showed no significant impact on bank profitability.

Jima (2017) argues that the net interest margin (NIM) of commercial banks varies across economies and seeks to identify the determinants of NIM in the Ethiopian banking industry, using a dataset from annual financial reports of Ethiopian banks and the National Bank of Ethiopia covering 1997–2014 Macroeconomic data, including GDP growth and inflation, come from the annual reports of the Ministry of Finance and Economic Development of Ethiopia, and expert opinions from banking professionals are incorporated to assess how internal and external changes affect bank performance The results show that cost efficiency, implicit interest payments, market competition, and economies of scale have a significant positive impact on NIM, while liquidity risk and poor management performance exert a significant negative impact By contrast, macroeconomic variables such as inflation and economic growth do not appear to significantly influence NIM.

Findings show that internal operational efficiency and robust business growth are both key determinants of performance for Ethiopia's commercial banks To optimize the industry-wide net interest margin (NIM), Ethiopian bank executives, advisors, and monetary authorities should prioritize strengthening these two pillars—enhancing operational efficiency and driving sustainable growth—so the entire banking sector can achieve higher profitability and resilience.

Sonia (2019) investigated the determinants of Net Interest Margin (NIM) in the

Using the Generalized Method of Moments (GMM) model, this study analyzes data from 426 banks across 15 European countries during 2013–2017 The model includes 11 independent variables: seven micro-level indicators—bank size (SIZE), capital (CAP), operational efficiency, liquidity (LIQ), credit risk, the ratio of non-interest income to operating revenue, and the ratio of non-performing loans to non-provisioned loans—and four macro-level variables—the Herfindahl index, GDP, inflation (INF), and the ratio of domestic credit to GDP The empirical results reveal that total asset size, operational efficiency, the non-interest income ratio, and the non-performing loans ratio have a statistically significant negative effect on net interest margin (NIM).

Harimurti (2022) used a multiple regression model to analyze the factors affecting the net interest margin (NIM) of 37 private banks in Indonesia, based on data collected from 2012 to 2021 The study found that the equity asset ratio (EAR), loan-to-deposit ratio (LDR), bank size, and operating costs and operating profit (OCOI) had a significant positive impact on NIM By contrast, non-performing loans (NPL) were negatively related to NIM, but not statistically significant In addition, external factors such as GDP growth and inflation also had a positive impact on NIM, but not statistically significant.

In Vietnam, the net interest margin (NIM) of commercial banks has drawn considerable academic attention Nguyen Minh Sang (2014) examined the determinants of NIM for Vietnamese commercial banks over 2008–2013, incorporating both micro factors and macroeconomic variables Using secondary data from 27 banks and quantitative methods, the study finds that liquidity, banking development level, the shareholding ratio, the loan-to-deposit ratio (LDR), and operating costs are positively associated with NIM Among macro factors, inflation is positively and statistically significantly related to NIM, while economic growth is negatively related The negative link with growth is attributed to the post-crisis period after the 2008–2013 global financial crisis, which eroded public trust in the financial system, prompting banks to raise loan rates to compensate for risk and thereby widen the NIM.

Hoang Vu Chinh (2017) applied the Generalized Method of Moments (GMM) estimation to identify factors affecting the net interest margin (NIM) of 27 Vietnamese commercial banks over 2006–2016 The analysis finds that outstanding loans and industry concentration are negatively correlated with NIM, while the loan-to-deposit ratio, liquidity risk, Lerner index, costs, capital risk, management efficiency, state bank reserve policy, outstanding loans, bank size, GDP, inflation rate, and the interest rate are positively correlated with NIM.

Pham Minh Dien and Duong Quynh Nga (2018) examined the impact of the

This study analyzes the determinants of net interest margin (NIM) for Vietnamese commercial banks over 2011–2015, incorporating the Herfindahl-Hirschman Index (HHI), the Lerner index, and the opportunity cost of the State Bank of Vietnam’s (SBV) reserve policy, alongside banking-sector factors such as operating costs, market share, bank size, and loan-to-deposit ratio (LDR) Using a panel data estimation with adjusted standard errors on a sample of 27 joint-stock commercial banks, the results show that the Lerner index, operating costs, and the opportunity cost of SBV reserves are positively correlated with NIM, while market share is negatively correlated with NIM In contrast, HHI and LDR have no statistically significant effect on NIM.

A 2020 study by Nguyen Duy Suu, Thu Quang Luu, Kim Hung Pho, and Michael McAleer analyzed 308 observations from 2008 to 2018 using a range of investigative methods to identify the factors influencing net interest margin (NIM) The empirical findings indicate that credit risk and operating costs are positively associated with NIM.

Net interest margin (NIM) in Vietnamese commercial banks is negatively affected by risk aversion, management quality, the deposit ratio, and business income These empirical findings help bank managers identify the key determinants of NIM and inform the development of targeted supervisory policies for Vietnamese commercial banks.

Nguyen Kim Chi and Nguyen Thi Minh Ngoc (2022) utilized panel data from the individual financial statements of 25 JSCB in Vietnam for the period from Quarter

From 2019 to Q1 2022, the Covid-19 pandemic significantly affected Vietnam’s national economy The study employs the Generalized Method of Moments (GMM) to examine factors influencing the net interest margin (NIM) of Vietnamese joint-stock commercial banks (JSCBs) The dataset excludes foreign banks, joint-venture banks, merged banks, banks without strategic shareholders, and banks with incomplete financial statements, while macroeconomic data such as inflation are drawn from the IMF The findings indicate that profitability, equity size, bank size, and operating cost ratios are positively and statistically significantly associated with NIM, whereas the liquidity ratio has a statistically significant negative effect on NIM Variables such as LOAN, INF, and GDP growth show no statistically significant impact on NIM in this study.

Table 2.4 1 Summary of Previous Empirical Studies

- Large banks in 7 countries: USA

Switzerland (94 banks), UK (32 banks), Italy (135 banks), France

- Opportunity cost of required reserves: (+)

-426 commercial banks in 15 European countries

- Non-interest income / Operating income: (−)

- Quy mô vốn chủ sở hữu (+)

HHI, and opportunity cost of reserves on the net interest margin of

- Opportunity cost of reserves at SBV: (+)

- Proportion of deposits and business income: (–)

"Factors affecting net interest margin during the COVID-

Vietnamese joint-stock commercial banks"

- Opportunity cost of reserves at SBV: (+)

Note: (+) indicates a positive correlation; (–) indicates a negative correlation; (0) indicates no observed correlation between the variables

Source: Compiled by the author

Research Gap

The dataset used in this thesis is updated through 2023, covering a 12-year period from 2012 to 2023, which enhances accuracy and data relevance compared with prior studies In addition, data were collected from multiple sources to ensure comprehensive coverage and reliability.

28 JSCBs in Vietnam, including the 4 leading state-owned CBs and 24 JSCBs, thereby ensuring both representativeness and diversity in the research sample

Although Vietnam's commercial banking system comprises diverse banks with different ownership structures, most previous studies fail to distinguish the unique roles and characteristics of state-owned commercial banks—such as Agribank, BIDV, Vietcombank, and Vietinbank—from joint-stock and foreign-owned banks Amid extensive restructuring from 2012 to 2023 and deepening international integration through major FTAs, state-owned banks have experienced more pronounced effects than other banks To capture this, we propose including a State Ownership (SOC) variable to clarify its specific influence.

To assess the impact of state-owned commercial banks on the net interest margin (NIM) of Vietnam's banking sector, the study incorporates the SOC variable into the research model to reveal the distinct effect of state ownership The dataset is updated to include the most recent information available as of 2023, enabling a direct comparison with previous related studies and strengthening the analysis of how ownership type influences NIM.

To establish a solid research foundation, this study reviews the basic theoretical framework of net interest margin (NIM), identifies the factors influencing NIM, and clarifies the relationships among these variables in Chapter 2 The micro-level determinants include bank size, equity capital, loan size, liquidity, credit risk, state bank reserve policy, and state ownership characteristics, while macroeconomic variables—economic growth and inflation—are incorporated into the model A comprehensive review of domestic and international studies highlights research gaps that guide variable selection for the model and shape subsequent chapters of the thesis.

Based on the fundamental theories and empirical findings summarized in Chapter

Chapter 3 develops a comprehensive research model to assess the impact of the selected variables on the NIM, and it outlines the research procedures and methodology employed in the thesis.

Here is the English academic-style translation of the section you provided from Chapter 3 of your thesis:

RESEARCH METHODOLOGY

Research Process

The primary objective of this thesis is to clarify how various factors affect the Net Interest Margin (NIM) of Vietnamese commercial banks (CBs) over 2012–2023 The study follows the structured methodology shown in Figure 3.1.1, guiding data collection, analysis, and interpretation to identify the drivers of NIM in Vietnam’s banking sector.

Review theoretical foundations and related empirical

Develop the research model and formulate research

Determine the research sample, collect and process research

Estimate the regression model using econometric

Test model selection and diagnose any violations in the

Test regression results against research hypotheses

Analyze and discuss results and provide managerial

Research Model

Building on theoretical foundations and prior empirical studies of the determinants of net interest margin (NIM) in both domestic and international contexts, this study identifies a set of key factors and develops a research model tailored to Vietnam The literature review spans studies from Europe, the Americas, the Middle East, and Southeast Asia, with Ong Tze San (2013) chosen as the primary reference model The selection rests on Malaysia—another Southeast Asian economy with similar conditions to Vietnam—and the finding that macroeconomic variables did not show statistically significant effects in that study Consequently, this thesis re-examines whether macroeconomic factors influence NIM in the Vietnamese context for the period 2012–2023.

The selected model incorporates nine explanatory variables affecting NIM Specifically:

NIM it : Ney Interest Margin of bank i at time t

SIZE it : Bank Size of bank i at time t

CAP it : Equity Capital of bank i at year t

LOAN it : Loan Size of bank of bank i at time t

LIQ it : Liquidity of bank i at time t

LLR it : Credit Risk of bank i at time t

SBRP it : State bank’s reserve policy of bank i at year t

SOC it : State Ownership Characteristics of bank i at year t

GDP t : Economic Growth at time t

INF t : Inflation at time t ԑ 𝒊 : Model Error, β 0 : Blocking coefficient.

Research Hypotheses

Net Interest Margin (NIM) is a widely used metric for assessing bank profitability, reflecting the efficiency of core intermediation activities such as deposit mobilization and lending Therefore, NIM is selected as the dependent variable in the research model Fungáčová and Poghosyan (2011) define Net Interest Margin as the growth rate of interest income relative to interest expenses, i.e., the percentage difference between total interest income and total interest expenses.

Bank Size (SIZE) influences the Net Interest Margin (NIM) of commercial banks, though findings remain inconclusive Hamadi and Awdeh (2012) state, "Bank size has a negative impact on NIM, as larger CBs may face higher risk exposure as they expand in scale." Conversely, Maudos and Guevara (2004) argue that "Larger banks are able to offer a more diversified range of products and services, which provides an advantage in raising funds at lower costs and earning higher profits compared to smaller CBs," suggesting that a larger bank size may enhance interest income by reducing interest-related costs.

H₁: SIZE is positively correlated with NIM

Equity Capital (CAP): Most studies have found a relationship between the size of equity capital and the net interest margin (NIM) In particular, Hamadi & Awdeh

Studies including 2012 and Saunders & Schumacher (2000) document a positive relationship between equity capital size and net interest margin (NIM) In Vietnam, Nguyen Kim Thu and Do Thi Thanh Huyen (2014) report similar results, indicating that a larger equity capital base provides banks with more internal funds for operations rather than relying on external funding sources, thereby reducing the cost of capital This reduction in funding costs lowers interest expenses and related costs, contributing to a higher NIM Moreover, a larger equity capital base mitigates operational risks and provides a cushion against bankruptcy Therefore, CAP is expected to have a positive impact on NIM.

H₂: CAP has a positive correlation with the NIM

Loan size drives lending, the traditional core activity and a primary income source for banks The profit derives from interest earned on loans after deducting the costs of mobilizing funds and the lending process As Hamadi and Awdeh (2012) state, "When the size of lending expands, the bank’s interest income also increases, thereby raising the NIM." In Vietnam, studies by Nguyen Thi Ngoc Trang and Nguyen Huu Tuan have examined related dynamics between loan size and bank performance.

(2015) yields consistent results with this view Therefore, the author expects a positive effect of the variable LOAN on NIM

H₃: LOAN has a positive correlation with the NIM

Liquidity (LIQ): "Although liquidity is currently one of the major concerns of

Liquidity risk, in certain circumstances, can still significantly affect the banking system (Đặng Văn Dân, 2015) Commercial banks often bolster liquidity by raising interest rates to attract deposits, a strategy that can reduce income and the net interest margin (NIM) As a result, banks incur higher opportunity costs to maintain liquidity at a stable level Research by Jima Meshesha Demie (2017) and Hamadi & Awdeh (2012) indicates an inverse relationship between liquidity and the NIM, implying a negative impact of liquidity on the NIM.

H₄: LIQ has a negative correlation with the NIM

Credit risk, reflected in loan loss reserves (LLR), is a core consideration for banks because lending generates most of their income but also carries the potential to impair performance It is the risk that borrowers may not fulfill their financial obligations as agreed at maturity As lending activity expands, banks face higher credit risk and must increase provisioning while also raising lending rates to compensate for the increased risk These higher rates and provisions raise borrowing costs and can squeeze the net interest margin (NIM) Multiple studies have documented a link between credit risk and NIM, including Angbazo (1997) and Maudos & Guevara.

(2004), Tarus (2012), and Hoang Vu Chinh (2017) Therefore, the author anticipates a positive effect of credit risk on the NIM

H₅: LLR has a positive correlation with the NIM

Under the State Bank's reserve policy, commercial banks must allocate a portion of customer deposits to a reserve account at the SBV, maintaining an average balance at least equal to the mandatory reserve level A higher reserve requirement raises the opportunity costs of holding these reserves, reducing banks' profitability and affecting interest income To maximize profits, banks seek higher interest income to offset these costs; consequently, an increase in the reserve requirement ratio tends to push the net interest margin (NIM) higher Saunders & Schumacher (2000) note that "as reserve requirements rise, banks will demand a higher net interest margin to compensate for the decline in profitability." In Vietnam, Pham Minh Dien (2018) found similar empirical results, leading to the expectation of a positive correlation between statutory reserves at the SBV and the NIM.

H₆: SBRP has a positive correlation with the NIM

State-owned commercial banks (SOC) typically operate more conservatively and exhibit lower risk appetites due to stringent regulatory oversight They prioritize capital preservation and risk minimization, often restricting lending to low-risk sectors such as consumer or secured loans rather than funding high-risk projects While this approach strengthens financial stability, it can dampen interest income growth and overall profitability because safer loans carry lower interest rates Empirical evidence, including Nguyen Thanh Binh et al (2022), shows a negative relationship between SOC status and net interest margin (NIM).

H₇: SOC has a negative correlation with the NIM

The commercial banking system acts as the backbone of the economy, so changes in overall economic conditions directly influence bank operations When economic growth accelerates, net interest income tends to rise, since higher growth typically brings increased personal income and stronger demand for financing for investment and production This linkage means that expansion phases often boost bank profitability through greater lending activity and wider interest margins, while slower growth can dampen credit demand and compress earnings.

(2013) argued that there is no significant relationship between these two variables Despite this, the author expects a positive correlation between economic growth and the Net Interest Margin

H₈: GDP has a positive correlation with the NIM

Inflation (INF): According to the International Monetary Fund (IMF), "Inflation affects the operations of CBs, including both deposit mobilization and lending activities" Nguyen Thi Ngoc Trang and Nguyen Huu Tuan (2015) also stated that

"Rising inflation contributes to higher NIM in CBs" Studies by Nguyễn Minh Sáng

(2014) and Hamadi & Awdeh (2012) similarly argue that inflation increases lending rates, thereby leading to a rise in NIM This is because when inflation trends upward, the central bank tends to implement monetary tightening policies, increasing lending interest rates Therefore, the author anticipates a positive correlation between inflation and Net Interset Margin

H₉: INF has a positive correlation with the NIM

Table 3.3 1 Summary of Research Hypotheses

Name Symbol Measurement Formula Expected

(Interest Income − Interest Expense) / (Total Interest- Earning Assets)

Bank Size SIZE ln (Total Assets) +

Capital Size CAP (Equity / Total Assets) ×

Fungáčová & Poghosyan (2011); Ong (2013); Saunders & Schumacher (2000); Hamadi (2012); Nguyễn Thị Ngọc Trang & Nguyễn Hữu Tuấn (2015); Nguyễn Kim Thu & Đỗ Thị Thanh Huyền (2014)

Loan Size LOAN (Total Loans / Total Assets) × 100% +

Hamadi (2012); Nguyễn Kim Thu & Đỗ Thị Thanh Huyền (2014)

Liquidity LIQ (Cash and Cash Equivalents

Fungáčová & Poghosyan (2011); Hamadi & Awdeh (2012); Jima (2017)

State bank’s reserve policy SBRP (Cash + Deposits at the

(Loan Loss Provisions / Total Customer Loans) ×

Dummy variable: value = 1 if state-owned (Agribank, BIDV, Vietcombank, Vietinbank); 0 otherwise

Growth Rate GDP Annual data collected from the World Bank + Hamadi (2012);

Inflation Rate INF Annual data collected from the World Bank +

Hamadi (2012); Nguyễn Minh Sáng (2014); Hoàng Trung Khánh & Vũ Thị Đan Trà (2015)

Research Methodology

This study analyzes Vietnamese commercial banks over a 12-year period from 2012 to 2023 Bank-level data are drawn from the balance sheets and income statements in the consolidated financial statements and annual reports published by Vietnamese commercial banks, which are considered highly reliable because they are audited by reputable audit firms In addition, macroeconomic indicators such as GDP growth and inflation rates were sourced from the World Bank’s official data portal (https://data.worldbank.org/).

Our data collection uses variables transparently and comprehensively disclosed by official sources for the period 2012–2023 The study focuses solely on Vietnamese commercial banks, excluding joint venture banks and foreign banks due to difficulties accessing their complete financial statements and the influence of exchange rates, which results in financial statement structures that differ from those of Vietnamese CBs Given data collection limitations, such as incomplete disclosures in the financial statements and annual reports of Vietnamese CBs, the final sample consists of 28 Vietnamese commercial banks.

This study uses panel data, combining cross-sectional and time-series observations, to examine the central banks' net interest margin (NIM) Because the NIM varies annually and is influenced by multiple factors, panel data analysis offers clear advantages, enabling more precise estimation of variable effects and reducing multicollinearity compared to purely cross-sectional or time-series approaches The data collection did not involve surveys; instead, data were entered into Excel and processed with Stata software version 17.0.

This study uses quantitative estimation methods to determine how bank size (SIZE), capital size (CAP), loan size (LOAN), liquidity (LIQ), credit risk (LLR), the State Bank’s reserve policy (SBRP), state ownership characteristics (SOC), inflation (INF), and the economic growth rate (GDP) influence the net interest margin (NIM) of Vietnamese commercial banks A panel data regression is employed to examine the impact of these nine independent variables on NIM, applying multiple estimation techniques to ensure robust, reliable results.

Descriptive statistics summarize a dataset by providing measures of central tendency and dispersion, notably the mean for central tendency and the maximum, minimum, and standard deviation for dispersion These core statistics reveal data patterns and variability, enhancing interpretation of the data and guiding the selection of appropriate analytical techniques for further analysis.

3.4.2.2 Correlation Analysis and Multicollinearity Testing

A correlation matrix quantifies the relationships between independent variables and the dependent variable, as well as among the independent variables themselves The correlation coefficients indicate whether relationships are positive or negative and whether they are strong or weak, helping identify meaningful associations This analysis also helps detect multicollinearity, a situation where independent variables are highly correlated, which can distort model estimates Multicollinearity is assessed using pairwise correlation coefficients and the variance inflation factor (VIF), providing a robust check on predictor independence.

This study conducts panel data analysis by sequentially applying three estimators: Ordinary Least Squares (OLS) to estimate the pooled regression model, followed by Fixed Effects Model (FEM) and Random Effects Model (REM) specifications to account for unobserved heterogeneity across cross-sectional units, thereby analyzing the collected panel data and comparing results to assess robustness.

(i) Pooled Ordinary Least Squares (Pooled OLS)

The pooled OLS model is specified as follows:

𝑌 𝑖𝑡 is the dependent variable for observation i in period t;

𝑋 𝑘𝑖𝑡 represents the independent variables for observation i in period t; α is the intercept term; β denotes the coefficient representing the effect of each independent variable; ԑ 𝑖𝑡 is the error term of the model

Pooled OLS is a simple regression technique that ignores spatial and temporal dimensions, assuming the impact of explanatory factors on NIM is the same across all CBs and constant over time In practice, this assumption is often unrealistic, as banks have unique characteristics and their effects on NIM can evolve over time A major drawback of Pooled OLS is its failure to account for unobserved heterogeneity, which can lead to unreliable estimation results (Phạm Thị Tuyết Trinh, 2016).

(ii) Fixed Effects Model (FEM)

The FEM is specified as follows:

Where: 𝑌 𝑖𝑡 is the dependent variable; 𝑋 𝑖𝑡 is the independent variable; α 𝑖 is the individual-specific intercept, 𝛽 is the slope coefficient; X; à 𝑖𝑡 is the error term

Compared to Pooled OLS, the fixed effects model (FEM) treats each bank as having its own time-invariant characteristics that can influence the dependent variable, with these individual effects potentially correlated with the explanatory variables By accounting for these fixed effects, FEM can provide more accurate estimates of how the independent variables impact NIM A key drawback is that FEM can exacerbate multicollinearity and it cannot estimate the effects of time-invariant variables.

(iii) Random Effects Model (REM)

The Random Effects Model (REM) treats bank-specific characteristics as random effects, modeling the bank intercept as α_i = α + u_i, where u_i is a bank-specific random component uncorrelated with the explanatory variables Unlike the Fixed Effects Model (FEM), which fixes the intercept and makes it time-invariant, REM allows α_i to vary across banks in a random way Substituting α_i = α + u_i into the standard panel data specification yields y_it = α + X_itβ + u_i + ε_it, where u_i captures between-bank heterogeneity and ε_it is the idiosyncratic error The key assumption is that u_i is uncorrelated with the regressors, enabling efficient estimation while accounting for unobserved bank differences in a panel data context.

Under the random-effects model (REM), Y_it is the dependent variable and X_it is the independent variable, with ε_i representing the individual cross-sectional effect and u_it the combined error term that captures both cross-sectional and time-series variation The REM treats the composite error term as part of the model and assumes that ε_i is uncorrelated with the regressors, allowing the use of both within- and between-entity variation Although REM can address some limitations of the fixed-effects model (FEM), its estimates rely on the assumption that ε_i is uncorrelated with X_it; if this assumption is violated, REM estimates become inconsistent and unreliable.

In summary, each model has its own assumptions, advantages, and limitations

From the preceding analysis, it is clear that both the fixed effects model (FEM) and the random effects model (REM) offer advantages over the pooled ordinary least squares (Pooled OLS) specification Nevertheless, to determine the most appropriate regression model for the study, the author will begin with the Pooled OLS approach in the thesis and then follow with the subsequent steps outlined below to guide model specification, comparison, and validation.

Step 1: Estimate the Pooled OLS model

Step 2: Estimate both the FEM and REM

Step 3: Perform the F-test to choose between the Pooled OLS and FEM models, with the following hypotheses:

𝐻 0 : : The Pooled OLS model is more appropriate

𝐻 1 : : The FEM model is more appropriate

If the p-value < 0.05 (at the 5% significance level), the null hypothesis 𝐻 0 : is rejected, and the FEM is selected; otherwise, the Pooled OLS model is retained

Step 4: After conducting the F-test:

+ If the Pooled OLS model is selected, perform the Breusch-Pagan Lagrangian

Multiplier (LM) test to decide between the Pooled OLS and REM models:

𝐻 0 : The Pooled OLS model is more appropriate

𝐻 1 : The REM is more appropriate

If the p-value < 0.05, reject 𝐻 0 : and select the REM; otherwise, retain the Pooled OLS model

+ If the FEM is selected, proceed to conduct the Hausman test to determine the suitability between FEM and REM, with the following hypotheses:

𝐻 0 : There is no correlation between the independent variables and the random error term (REM is appropriate)

𝐻 1 : : There is correlation between the independent variables and the random error term (FEM is appropriate)

If the p-value < 0.05, reject 𝐻 0 : and choose the FEM; otherwise, select the REM

After estimation, diagnostic tests are conducted to evaluate model assumptions and validity These include checks for multicollinearity, heteroskedasticity, and autocorrelation:

Multicollinearity occurs when two or more independent variables are highly correlated It is first assessed using correlation coefficients; a correlation coefficient ≥ 0.8 may indicate multicollinearity, though this may not always be reliable, as multicollinearity can exist even with lower correlations Therefore, the variance inflation factor (VIF) is also used; a VIF ≥ 10 suggests serious multicollinearity that can distort regression results In such cases, highly correlated variables should be replaced with more appropriate alternatives (Gujarati, 2004).

Heteroskedasticity occurs when, after regression analysis, the variance of the error terms (ε) or the estimated residuals is not constant and does not follow a random distribution This violates the classical linear regression assumption of homoscedasticity, which requires constant error variance for the model to be valid When heteroskedasticity is present, ordinary least squares (OLS) remains unbiased and consistent but is no longer the best linear unbiased estimator (BLUE) Consequently, the statistical inferences drawn from the model can become unreliable and may lead to incorrect conclusions.

Heteroskedasticity testing is typically performed after the Hausman test, with two scenarios depending on the chosen panel model: if the Random Effects Model (REM) is selected, the xttest0 command is used to test for heteroskedasticity; if the Fixed Effects Model (FEM) is chosen, the xttest3 command is applied The null hypothesis states homoskedasticity (constant error variance), while the alternative indicates heteroskedasticity across observations These steps ensure the diagnostic test aligns with the specified model and helps confirm the reliability of the panel data analysis.

• H₀: Homoskedasticity is present (error variance is constant)

• H₁: Heteroskedasticity is present (error variance is not constant)

RESEARCH RESULTS AND DISCUSSION

Descriptive Statistics Results

The panel data were collected by the author from 28 Vietnamese commercial banks over the period 2012–2023 The descriptive statistical indicators are shown in the following table:

(Source: Author’s extraction from Stata 17 software)

(Refer to Appendix 1.1 for details)

Table 4.1.1 presents the descriptive statistics results, in which both the dependent and independent variables have the same number of observations (336), collected from

28 Vietnamese commercial banks over the 12-year period from 2012 to 2023

Net Interest Margin (NIM) is the dependent variable in the model, ranging from a low of 4.4% recorded by Ho Chi Minh City Development Joint Stock Commercial Bank (HDBank) in 2013 to a high of 8.84% observed at Vietnam Prosperity Joint Stock Commercial Bank (VPBank) in 2019 Across Vietnamese commercial banks in this study, NIM values do not vary widely, fluctuating between 0.044 and 0.0884, with no negative values, signaling a relatively stable level of profitability from credit activities The standard deviation of 0.0121 around a mean of 0.0296 indicates low dispersion around the average, meaning banks, on average, earned 2.96% net interest income per unit of their interest-earning assets from 2012–2023 However, not all banks reached this average during the period; some achieve higher NIMs by targeting high-margin lending segments such as consumer lending, while others with more conservative asset structures or higher funding costs may report lower NIMs.

Bank size (SIZE) is the independent variable with the highest standard deviation

Among all variables in the model, bank size shows significant variation relative to the market average, reflecting the Vietnamese banking system’s mix of long-established state-owned banks and newly established private commercial banks The average SIZE value is 32.6592, with the highest SIZE value on record in 2023 belonging to the Joint Stock Commercial Bank for Investment and Development of Vietnam (BIDV) at 35.3721, and the lowest at 30.2175 recorded by Baoviet Joint Stock Commercial Bank (Baoviet Bank) in 2012 In general, banks have increasingly expanded their scale of operations to reach a broader customer base and strengthen market credibility, indicating an upward trend in bank size over time.

Equity Capital (CAP) shows a sample mean of 0.0877 with a standard deviation of 0.0361 In 2022, Vietcombank (VCB) posted the lowest CAP ratio at 0.0293, despite its large equity base, as substantial total assets kept the ratio relatively low; this pattern suggests heavy reliance on capital mobilization and the use of high financial leverage to maximize profitability By contrast, the highest CAP ratio occurred in 2013 at 0.2384 for SaigonBank (SGB), where equity comprised nearly a quarter of total assets Such high CAP levels are viewed as desirable for Vietnamese commercial banks, supporting stronger liquidity through core capital, signaling solid financial health, and providing a buffer against business risks.

Loan size (LOAN) shows a sample mean of 0.5860 with a relatively high standard deviation of 0.1136, indicating substantial variability in credit activity across Vietnamese commercial banks This variability reflects differing strategic focuses: some banks prioritized large-scale lending such as corporate loans and infrastructure financing, while others concentrated on financial services and consumer lending LOAN ratios span from a low of 0.2162 recorded by SeABank (SSB) in 2012 to a high of 0.7881 posted by BIDV, underscoring the diversity of lending practices within the Vietnamese banking sector.

From 2012 to 2023, the LOAN variable demonstrated a clear upward trend, indicating that commercial banks have effectively fulfilled their financial intermediation role by expanding lending activities and meeting the economy’s rising capital demands.

Liquidity (LIQ) has an average value of 0.1779 and a standard deviation of

According to the descriptive statistics table, the highest LIQ value was observed in 2012 at 0.5211 for Southeast Asia Commercial Joint Stock Bank (SeABank – SSB), indicating that over 50% of the bank’s assets that year were held in cash or cash equivalents Conversely, the lowest LIQ value was 0.0452, recorded by Saigon Thuong.

Liquidity is a constant concern for Sacombank (STB) and other Vietnamese commercial banks, as it reflects their ability to meet customers’ immediate cash withdrawal and disbursement demands At the same time, banks must maintain liquidity at an optimal level to maximize capital utilization while ensuring financial safety and stability In 2017, Sacombank emphasized balancing liquidity management with prudent capital deployment to support reliable service and sustainable growth within Vietnam’s banking sector.

Credit risk (LLR) has a mean value of 0.0141 and a standard deviation of

With a standard deviation of 0.0051, this variable ranks among the smallest in the model, signaling that commercial banks have implemented effective policies to curb non-performing loans in their lending activities The lowest LLR value observed was 0.0067 at Nam A Commercial Joint Stock Bank (NamABank) in 2013, while the highest was 0.0338 at the Vietnam Bank for Agriculture and Rural Development (Agribank) in 2012.

State Bank’s Reserve Policy (SBRP) shows a standard deviation of 0.0199 and a mean of 0.0373, aligning with the State Bank of Vietnam’s (SBV) typical reserve requirements for commercial banks during the research period The data also reveal a maximum of 0.1245 in 2023 for SaigonBank – Saigon Bank for Industry and Trade (SaigonBank) and a minimum of 0.0080 recorded by Viet A Commercial Joint Stock Bank (VietABank) in 2017, indicating a relatively wide fluctuation range around the overall sample mean.

State Ownership Characteristics (SOC) has a sample mean of 0.1429 and a standard deviation of 0.3504 The variable was constructed by the author, with state-owned banks such as BIDV, Vietcombank, VietinBank, and Agribank assigned a value of 1, while the remaining 24 joint stock commercial banks were assigned a value of 0.

As this is a group-specific dummy variable, the author does not provide detailed descriptive statistics for it in this section

Vietnam’s GDP growth averaged 0.0516 with a standard deviation of 0.0171 over the 2012–2023 period, reflecting a generally stable yet variable economic expansion The downturn during the COVID-19 era saw a trough of 0.0258 in 2021, followed by 0.0291 in 2020, underscoring the pandemic’s impact on production, consumption, and livelihoods By comparison, the peak growth rate was 0.0708 in 2018, marking the strongest expansion since the global financial crisis of 2008.

Inflation (INF) has a mean value of 0.0362, which is consistent with the State

Bank of Vietnam targets inflation below 4%, a goal that anchors price stability and supports sustainable growth The standard deviation of INF is 0.0214, a relatively low figure indicating that government efforts to maintain inflation at a stable and controlled level have reduced volatility and fostered a predictable economic environment In the 2012–2023 period, the highest inflation rate reached 0.0909, illustrating the dynamics of price pressures within the period while underscoring the overall effectiveness of inflation management.

2012, and the lowest was 0.0063 in 2015.

Correlation Analysis

NIM SIZE CAP LOAN LIQ LLR SBRP SOC GDP INF

Note: ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively

Table 4.2.1 presents the correlation analysis between the dependent variable NIM and the independent variables, as well as the correlations among the independent variables themselves The correlation coefficients reveal whether these relationships are positive or negative and indicate their strength, with the linear correlation coefficient valid within the range of -1 to 1.

Equity Capital (CAP) has the strongest positive association with Net Interest Margin (NIM), with a correlation coefficient of 0.376 Loan Size (LOAN) follows, showing a strong positive relationship with NIM at 0.251 Additionally, Bank Size (SIZE), Credit Risk (LLR), State bank’s reserve policy (SBRP), and Inflation (INF) are all positively correlated with NIM, indicating that these variables tend to rise together with Net Interest Margin.

Conversely, Liquidity (LIQ), State Ownership Characteristics (SOC), and Economic Growth (GDP) exhibit negative correlations with NIM In particular, Liquidity (LIQ) is negatively correlated with NIM at a high level of statistical significance (1%), while Economic Growth (GDP) shows a negative relationship with NIM at the 10% significance level

The results from Table 4.2.1 indicate that the absolute values of all correlation coefficients in the matrix are less than 0.8 and fall within the range of -1 to 1 Therefore, based on the benchmark proposed by Farrar & Glauber (1967), multicollinearity is not present in the model However, this assumption may not be entirely accurate, as multicollinearity can still occur even when correlation coefficients are low Hence, the author proceeds to use the Variance Inflation Factor (VIF) to test for multicollinearity in the model The results of this test are presented as follows:

Based on the results in Table 4.2.2, all VIF values in the model are less than 10, ranging from 1.1 to 3.1 According to the threshold proposed by Bahiru Workneh

According to a 2014 guideline, a VIF value of 10 or higher signals severe multicollinearity In this study, the average VIF value is 1.78, which is below 2 Therefore, based on the VIF values, multicollinearity is not present in the research model.

Selection of the Appropriate Model

After performing descriptive statistics, constructing the correlation matrix, and testing for multicollinearity, the study proceeds to panel data regression using three traditional estimation methods: Pooled Ordinary Least Squares (Pooled OLS), Fixed Effects (FE), and Random Effects (RE) Each method offers a different lens on the data—Pooled OLS pools all observations and ignores unobserved heterogeneity, FE accounts for time-invariant characteristics within entities, and RE treats the unobserved effects as random and uncorrelated with the regressors—allowing a comparison of coefficient estimates and model fit across specifications to identify the most reliable relationships.

Using both the Fixed Effects Model (FEM) and the Random Effects Model (REM), this study examines the extent to which the independent variables influence the Net Interest Margin (NIM) All regression analyses are conducted with Stata 17.0, enabling a robust comparison of FEM and REM estimates.

Regression results for the Pooled OLS, Fixed Effects (FEM), and Random Effects (REM) models are reported in Appendices 1.3, 1.5, and 1.6, respectively, and Table 4.3.1 provides a consolidated summary of the findings across all three specifications.

Table 4.3 1 Summary of Regression Results for Pooled OLS, FEM, REM Models

Note: ***, **, * denote statistical significance at the 1%, 5%, 10% levels, respectively

(Source: Author’s extraction from Stata 17 software) (Refer to Appendix 1.3, 1.5 and 1.6 for details)

Table 4.3.1 presents the regression results for the Pooled OLS, Fixed Effects Model (FEM), and Random Effects Model (REM) The analysis shows moderate explanatory power across the models, with R-squared values of 48.3% for Pooled OLS, 41.2% for FEM, and 41.1% for REM Consequently, the independent variables explain 48.3% of the variation in the Net Interest Margin (NIM) under the Pooled OLS specification, compared with 41.2% (FEM) and 41.1% (REM).

Accordingly, the Pooled OLS and REM models each report 7 statistically significant variables, while the FEM model has 4 statistically significant variables, detailed as follows:

• Pooled OLS model: The regression results show that the variables Bank Size

Our analysis shows that SIZE, Capital Adequacy (CAP), Liquidity (LIQ), Loan Ratio (LOAN), State Ownership Characteristics (SOC), and Inflation (INF) are all positively correlated with net interest margin (NIM) at the 1% significance level In addition, Credit Risk (LLR) has a positive impact on NIM at the 5% significance level By contrast, State Bank’s Reserve Policy (SBRP) and the Economic Growth Rate (GDP) do not exhibit statistical significance with respect to NIM.

Within a Fixed Effects Model (FEM), SIZE, CAP, LOAN, and INF exhibit a positive, statistically significant association with NIM at the 1% significance level, highlighting these variables as robust predictors of net interest margins Conversely, LIQ, LLR, SBRP, SOC, and GDP show no statistically significant relationship with NIM.

• Random Effects Model (REM): The variables SIZE, CAP, LOAN, SOC, and

INF is statistically significant at the 1% level; LIQ has a positive effect on NIM at the 10% significance level, while LLR, SBRP, and GDP do not show statistical significance, with GDP and SBRP remaining insignificant across all three models.

Panel data regression is estimated using three specifications—Pooled OLS, the Fixed Effects Model (FEM), and the Random Effects Model (REM) The author then applies statistical tests to compare these estimators and determine the most appropriate model for the study, ensuring robust and efficient inference about the relationships of interest The model selection balances bias and variance, identifying the specification that best captures the data-generating process in the context of the research question.

First, as mentioned in Section 3.4.2, the author performs the F-test to evaluate and compare the Pooled OLS and FEM models with the following hypotheses:

• H₀: The Pooled OLS model is more appropriate

• H₁: The Fixed Effects Model (FEM) is more appropriate

F-test results in Table 4.3.2 show a p-value of 0.0000, below the 0.05 significance level, which leads to rejection of the null hypothesis (H0) Therefore, the Fixed Effects Model (FEM) is selected for further analysis.

Next, the author performs the Breusch-Pagan Lagrangian Multiplier (LM) test to compare the Pooled OLS and Random Effects Model (REM) with the following hypotheses:

• H₀: The Pooled OLS model is more appropriate

• H₁: The Random Effects Model (REM) is more appropriate

Table 4.3 3 Breusch-Pagan Lagrangian Test Results

H0: sigma(i)^2 = sigma^2 for all i Chi2 (28) = 1009.48

Breusch-Pagan Lagrange Multiplier test results (Table 4.3.3) yield a p-value of 0.0000, which is below the 0.05 significance threshold, providing evidence to reject the null hypothesis Therefore, the Random Effects Model (REM) is selected for further analysis.

Subsequently, the author conducts the Hausman test to evaluate and choose between the Fixed Effects Model (FEM) and the Random Effects Model (REM), with the following hypotheses:

• H₀: There is no correlation between the independent variables and the random error terms (REM is appropriate)

• H₁: There exists a correlation between the explanatory variables and the random error terms (FEM is appropriate)

Test of H0: Difference in coefficients not systematic chi2(9) = (b-B)'[(V_b-V_B)^(-1)](b-B) = 19.17 Prob > chi2 = 0.0140

According to the results of the Hausman test in Table 4.3.4, the p-value = 0.0140

< 0.05 Therefore, there are sufficient grounds to reject the null hypothesis (H₀), and the Fixed Effects Model (FEM) is selected for further analysis

Conclusion: After performing model selection tests among the three models—

Pooled OLS, FEM, and REM—through the F-test, Breusch-Pagan Lagrangian test, and

Hausman test, the Fixed Effects Model (FEM) is assessed to be the most appropriate model for this study

4.3.3 Testing for Violations of the FEM Assumptions

After the Hausman test, heteroskedasticity is evaluated using the Breusch-Pagan Lagrangian Multiplier test Since a fixed effects model (FEM) is selected, Stata 17’s xttest3 command is employed to test for heteroskedasticity, with the null hypothesis H0: homoskedasticity and the alternative H1: heteroskedasticity.

• H₀: The variance of the error terms is constant (homoskedasticity)

• H₁: The variance of the error terms is not constant (heteroskedasticity)

H0: sigma(i)^2 = sigma^2 for all i chi2 (28) = 1009.48 Prob>chi2 = 0.0000

According to Table 4.3.5, the Breusch-Pagan Lagrangian Multiplier test produced a p-value of 0.0000, which is less than 0.05 This result leads to the rejection of the null hypothesis of homoskedasticity, indicating the presence of heteroskedasticity in the FEM model. -**Support Pollinations.AI:** -🌸 **Ad** 🌸Powered by Pollinations.AI free text APIs [Support our mission](https://pollinations.ai/redirect/kofi) to keep AI accessible for everyone.

To test for autocorrelation, the author conducts the Wooldridge test using the xtserial command in Stata 17, with the following hypotheses:

Wooldridge test for autocorrelation in panel data H0: no first-order autocorrelation

Results from the Wooldridge test, as shown in Table 4.3.6, yield a p-value of 0.0000, which is well below the 0.05 significance threshold; thus, the null hypothesis of no autocorrelation is rejected, indicating the presence of autocorrelation in the FEM (fixed effects model).

In summary, the Fixed Effects Model (FEM) exhibits both heteroskedasticity and autocorrelation To address these violations, the analysis uses panel data regression estimated with Feasible Generalized Least Squares (FGLS) to identify the most optimal model for the study.

4.3.4 Regression Using the FGLS Method:

The regression is performed using the Feasible Generalized Least Squares (FGLS) method to overcome the issues of heteroskedasticity and autocorrelation in the model

Table 4.3 7 Regression Results Using FGLS Model

NIM Intercept Standard Error t-statistic P-value

Note: ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively

Table 4.3.7 shows that the FGLS regression yields statistically significant results at the 1% level (Prob > chi2 = 0.0000), indicating the model is appropriate for estimating NIM Among the nine predictors, SIZE, CAP, LOAN, SBRP, and INF have positive associations with NIM at the 1% significance level, while LIQ and SOC have negative associations with NIM at the same level GDP is positively related to NIM but at a lower significance level (10%), and LLR is not statistically significant, implying its impact on NIM is not supported by the data. -**Support Pollinations.AI:** -🌸 **Ad** 🌸Powered by Pollinations.AI free text APIs [Support our mission](https://pollinations.ai/redirect/kofi) to keep AI accessible for everyone.

4.3.5 Regression using the GMM method

Discussion of Research Findings

In developing the model and hypotheses, the researcher identified nine independent variables, including banking-specific characteristics and macroeconomic factors expected to influence the Net Interest Margin (NIM) According to Table 4.4.1, GMM regression results show that five variables have a statistically significant impact on NIM: Bank size (SIZE), Equity Capital (CAP), Loan Size (LOAN), Credit Risk (LLR), and Economic growth (GDP).

Conversely, the variables Liquidity (LIQ), State bank’s reserve policy (SBRP), State Ownership Characteristics (SOC), and Inflation (INF) were found to have no statistically significant effect on the NIM

Table 4.4 1 Summary of Empirical Results Compared to Research Hypotheses

Expected Sign Result Significance Level

Note: (+) indicates a positive correlation; (–) indicates a negative correlation

Bank size has a positive impact on net interest margin (NIM) at the 5% significance level during 2012–2023, with a 1% increase in bank size associated with a 70.2% rise in NIM (ceteris paribus) This result aligns with the authors’ expectations and prior studies by Maudos & Guevara (2004), Fungáčová & Poghosyan (2011), and Hamadi (2012) Consequently, H1: SIZE has a positive correlation with NIM is accepted.

Figures 4.4.1 The correlation between NIM and SIZE

Capital size shows a positive relationship with the net interest margin (NIM) at the 10% significance level during the 2012–2023 period Specifically, a 1% increase in capital size is associated with a 1.28% increase in NIM, holding other factors constant This finding aligns with the authors’ expectations and is supported by existing studies, including Fungáčová & Poghosyan.

(2011), Ong (2013), Saunders & Schumacher (2000), Hamadi (2012), Nguyen Thi Ngoc Trang & Nguyen Huu Tuan (2015), and Nguyen Kim Thu & Do Thi Thanh Huyen

(2014) The hypothesis H₂: CAP has a positive correlation with NIM is accepted

Figure 4.4.2 The correlation between NIM and CAP

Across the 2012–2023 period, loan size exhibits a statistically significant positive impact on net interest margin (NIM) at the 1% significance level Specifically, a 1% increase in loan size is associated with a 0.09% rise in NIM, holding other factors constant This result aligns with the author’s expectations and with prior research by Hamadi (2012) and Nguyễn Kim Thu & Đỗ Thị Thanh Huyền (2014) Therefore, Hypothesis H3, positing a positive correlation between LOAN and NIM, is accepted.

Figure 4.4.3 The correlation between NIM and LOAN

Liquidity does not have a statistically significant effect on Net Interest Margin (NIM), and the analysis provides insufficient evidence to determine any impact of liquidity on NIM or to accept hypothesis H4, which posits a negative correlation between LIQ and NIM.

Credit risk has a statistically significant positive impact on the net interest margin (NIM) at the 5% level during the period 2012–2023 The correlation indicates that a 1% increase in credit risk is associated with a 29% increase in NIM, ceteris paribus This outcome aligns with the author's expectations and with prior studies such as Nguyen Minh Sang (2014) and Hoang Vu Chinh (2017) Consequently, hypothesis H5, which posits that CR has a negative correlation with NIM, is accepted.

Figure 4.4.4 The correlation between NIM and LLR

4.4.6 State Bank’s Reserve Policy (SBRP)

The State Bank's reserve requirement (SBRP) does not show statistical significance with respect to net interest margin (NIM) Consequently, there is insufficient evidence to confirm any impact of the reserve requirement on NIM or to accept hypothesis H6, which posits a positive correlation between SBRP and NIM.

State ownership characteristics do not exhibit a statistically significant relationship with NIM Consequently, the study finds insufficient evidence to support the impact of SOC on NIM or to accept hypothesis H7, which states that SOC has a positive correlation with NIM.

Between 2012 and 2023, GDP growth has a positive and statistically significant impact on the net interest margin (NIM) at the 10% level The analysis shows that a 1% increase in economic growth is associated with a 6.69% increase in NIM, with all else equal This finding is consistent with existing literature that links higher economic activity to wider net interest margins and improved bank profitability.

NIM LLR author's expectations and prior research such as Hoàng Vũ Chính (2017) and Hamadi

(2012) Thus, hypothesis H₈: GDP has a positive correlation with NIM is accepted

Figure 4.4.5 The correlation between NIM and GDP

Inflation (INF) does not have a statistically significant effect on Net Interest Margin (NIM) Consequently, there is insufficient basis to determine INF's impact on NIM or to support Hypothesis H9, which posits a positive correlation between INF and NIM.

Following the methodology outlined in Chapter 3, this study begins with descriptive statistics to report the mean, standard deviation, minimum, and maximum values of the research variables It then assesses the correlations among variables and checks for multicollinearity To identify the most appropriate model, the analysis estimates pooled OLS, fixed effects (FEM), and random effects (REM) specifications The chosen model is subsequently tested for violations such as autocorrelation and heteroskedasticity, and these limitations are addressed through robust standard errors and other corrective procedures as needed.

Employing the Feasible Generalized Least Squares (FGLS) approach and subsequently the Generalized Method of Moments (GMM) to address endogeneity, the regression results indicate that five of the nine independent variables proposed in Chapter 3—SIZE, CAP, LOAN, LLR, and GDP—have statistically significant effects on NIM, while SOC, SBRP, and INF show no statistically significant impact.

Building on the findings reported in Chapter 4, Chapter 5 presents managerial implications designed to improve the net interest margin (NIM) of Vietnamese commercial banks It translates these findings into practical strategies for asset-liability management, pricing, product mix, and risk controls that banks can implement to strengthen NIM performance The chapter also discusses the study's limitations, including data constraints and generalizability, and outlines directions for future research to deepen understanding of NIM determinants in Vietnam's banking sector Together, these sections bridge the empirical results with actionable recommendations and a roadmap for ongoing scholarly inquiry.

CONCLUSION AND IMPLICATIONS

Tiêu đề	Factors affecting the net interest margin of Vietnamese commercial banks
Tác giả	Dang Nguyen Phuong Linh
Người hướng dẫn	Prof. Dr. Dang Van Dan
Trường học	Banking University of Ho Chi Minh City
Chuyên ngành	Finance - Banking
Thể loại	Master thesis
Năm xuất bản	2025
Thành phố	Ho Chi Minh City

Định dạng
Số trang	109
Dung lượng	1,51 MB