The factors affect to the non performing loans of commercial banks in vietnam

INTRODUCTION

The urgency of the study

The banking sector is crucial for a country's economy, acting as a bridge between surplus and deficit capital Banks efficiently utilize depositor funds, share risks, and significantly contribute to economic growth, making them central to the financial system and often at the heart of financial crises (Franklin and Elena, 2008) Financial institutions are essential for converting deposits into productive investments, thereby operating the entire economy (Podder and Mamun, 2004) Consequently, the stability of the banking system is vital, with the non-performing loan (NPL) ratio being a key factor influencing financial instability and potential banking crises.

Rising non-performing loans lead to the collapse of the banking system and affect the entire economy as well as a country‘s politics Speacifically, the NPL crisis in

From 2007 to 2012, Vietnam faced significant challenges due to the global financial crisis, leading to an increase in non-performing loans (NPLs) Although effective bad debt settlement policies implemented by the state helped maintain control over the NPL ratio by 2019, the onset of the Covid-19 pandemic at the end of that year disrupted economic stability and altered global financial policies Consequently, the NPL ratio began to fluctuate, resembling the crisis levels experienced in 2012 To prevent a banking crisis and mitigate the economic downturn following the pandemic, it is crucial for bank managers to identify the factors influencing bad loans and develop effective strategies for managing this risk.

The significant financing costs associated with non-performing loans (NPLs) are primarily managed by state-controlled asset management enterprises, which focus on resolving bad debts from financial institutions This process adversely affects government budget revenues, with banks' NPL settlement costs representing 10% to 20% of the country's total GDP (Boudriga et al., 2009) Consequently, the management and minimization of financial costs related to non-performing loans has become a focal point for economic researchers over the years.

According to SBV data, the internal bad debt situation of the whole banking industry in Vietnam in 2012 – 2022 fluctuated quite a lot The peak NPL ratio in

The bad debt ratio in Vietnam rose to 4.86% in 2012 due to the global financial crisis and monetary easing policies from 2006 to 2007, along with rapid credit growth in 2009-2010 However, effective management policies led to a decline in bad debt, reaching 2.46% by 2016 This downward trend continued, with bad debts dropping to 1.6% in 2019 The COVID-19 pandemic in 2020 caused a reversal, increasing bad debt to 1.7%, which rose to 1.9% in 2021 when considering probable bad debts To support customers affected by the pandemic, the government implemented Circulars No 03/2021 and 14/2021 to ease debt repayment Without these measures, bad debts could have surged to 8.2% Economic analysts anticipated fluctuations in bad debt ratios for 2022, and financial reports indicated an upward trend, with notable increases at several banks: VPBank at 4.78%, Saigonbank from 1.97% to 2.12%, and TPBank from 2.34% to 2.88% Despite slight increases in small and medium-sized banks, seven banks maintained bad debt ratios below 1%, and two banks achieved a provision ratio for bad loans exceeding 300%.

Chart 1.1 NPL ratio in the period 2012 – 2021 (% of total outstanding loans)

Source: The State Bank of Vietnam (SBV)

Previous research on non-performing loans has explored various factors influencing bad debts in Vietnamese commercial banks during different economic conditions, particularly during periods of stability or recession linked to financial crises However, these studies have not accounted for the impact of epidemic situations, such as those experienced in recent years.

2019 to 2021 Therefore, the author tends to carry out the study "THE FACTORS

The article "Affect to the Non-Performing Loans of Commercial Banks in Vietnam" explores both internal banking and macroeconomic factors, offering diverse insights for managers The research findings present a new perspective on non-performing loans and suggest actionable solutions for bank management to address this critical issue effectively.

Research objectives

This thesis aims to analyze the factors influencing non-performing loans (NPL) in Vietnam's Joint Stock Commercial Banks Based on the research findings, the study offers several recommendations to mitigate the risk of bad debt in these banks.

To achieve the general goal, the research needs to achieve specific goals following:

 Firstly, Identify the factors affecting the NPLs of Vietnamese joint stock commercial banks From there, the author builds appropriate research models

 Secondly, measure the direction and level of influence of factors affecting the NPLs based on the result of the study

 Lastly, the results suggest some measures to help limit and control the non- performing loans of Vietnam Joint Stock Commercial Banks.

Research Questions

The author fabricates a list of the following key questions to accomplish the research objectives:

 What factors impact the NPLs of Vietnamese commercial banks during the period 2012 - 2021?

 What direction do the factors in the research model impact the NPLs?

 From the research results obtained, what policy implications does the author propose to improve the NPLs situation?

Subject and scope of research

This study examines the factors influencing the Non-Performing Loan (NPL) ratio of 28 commercial banks in Vietnam The selection of these banks was based on the availability of transparent and consistent data throughout the research period.

To ensure data reliability, the author selects a sample of 28 joint stock commercial banks in Vietnam, including ABB, ACB, AGRI, BIDV, BAOVIETBANK, BVB, CTG, EIB, HDB, KLB, MBB, MSB, NAB, OCB, PGB, PVCOMBANK, SCB, SGB, NVB, SHB, SSB, STB, VCB, TCB, TPB, VAB, VPB, and VIB Financial data is compiled from the banks' financial statements and annual reports, while macroeconomic data is sourced from the General Statistics Office, the State Bank, and the Federal Reserve Economic Data.

The study analyzed data from 2012 to 2021, encompassing a complete economic cycle that included a post-depression phase from 2012 to 2015, a development stage from 2016 to 2019, and a global recession from 2019 to 2021 This timeframe captures all critical stages of economic progression, from recession to recovery, followed by growth and eventual decline, making it an ideal period for research.

Research methods and data

The article uses a combination of both quantitative and qualitative methods

The author begins by synthesizing fundamental theories of bad debt and economic principles to establish a foundation for selecting model variables Subsequently, they summarize and compile statistical data, analyze, evaluate, and compare findings from previous studies to develop a comprehensive research model.

This article emphasizes the importance of quantification in analyzing bad debt, utilizing Excel and Stata for data synthesis and statistical analysis Key methodologies include Ordinary Least Squares (OLS), Fixed Effects Model (FEM), and Random Effects Model (REM) for model measurement, alongside F-Test and Hausman tests for model comparison and accuracy To ensure the robustness of the chosen model, Wooldridge and Modified Wald tests will be conducted to assess autocorrelation and variance Should any issues arise from these tests, the SGMM method will be employed to address the resulting errors effectively.

This study analyzes financial statements from 28 listed commercial banks in Vietnam over a decade (2012-2021), utilizing data sourced from the Fiin Pro platform, renowned for its comprehensive financial insights Additionally, macroeconomic data is gathered from the General Statistics Office and the State Bank of Vietnam, while the Economic Policy Uncertainty ratio is derived from the Federal Reserve Economic Data.

Meaning of the topic

Non-performing loans (NPLs) serve as a crucial indicator of risk within the financial sector, making them a focal point for economic researchers An increase in NPLs not only disrupts the banking industry but also has broader implications for the overall economy Given the unique business model of banks, which act as key conduits for capital flow, the ratio of non-performing loans significantly influences a nation's economic health This study aims to provide empirical evidence regarding the factors affecting non-performing loans in listed commercial banks in Vietnam from 2012 onwards.

This study examines the influence of internal bank factors—such as previous non-performing loans, provisioning costs, operating expenses, bank leverage, size, and profit—alongside macroeconomic determinants like inflation, GDP growth rate, and the World Uncertainty Index By incorporating economic policy uncertainty, it aims to assess how policymaking affects non-performing loans in Vietnam The research enhances the understanding of macroeconomic variables, providing valuable insights for bank managers and economic researchers to reference, expand upon, and apply in future studies.

The article discusses the management implications of non-performing loans (NPLs) derived from research findings, emphasizing the influence of various internal factors on bad debt control within banks It offers strategic solutions and forecasts for the NPLs of commercial banks in Vietnam, aiming to support economic growth amid the ongoing global recession.

Layout of the thesis

The structure of the graduation thesis comprises 5 chapters:

Introduction to the reasons for choosing the topic, research objectives, research questions, subjects and scope of research, research methods and data, and topic meaning

CHAPTER 2: THEORETICAL FOUNDATIONS AND EMPIRICAL STUDIES

This article explores the theoretical foundations of factors influencing Non-Performing Loans (NPLs) in commercial banks, providing a comprehensive overview of the NPL concept, measurement methods, and their broader economic implications Additionally, Chapter 2 reviews relevant theories and prior empirical research to pinpoint gaps in existing studies, setting the stage for future research expectations.

CHAPTER 3: RESEARCH MODELS AND METHODS

This chapter covers the research process, research models, and research methods The author provides the basis for model proposals, research hypotheses, and sample selection bases for collecting research data

The article presents regression analysis results and conducts key tests to derive experimental findings The author evaluates the hypotheses formulated from the regression outcomes In the discussion section, the author highlights the final results and projects the trend of bad debt fluctuations using a linear regression model.

Conclude and make recommendations to policymakers, management agencies, the State and commercial banks The author lays out the limitations of the study and the direction of future research development

This chapter outlines the research's purpose, objectives, scope, and structure of the graduation thesis It offers readers a concise overview of the study, allowing them to quickly identify key information at a glance.

CHAPTER 2: THEORETICAL FOUNDATIONS AND EMPIRICAL

Theoretical basis of non-performing loans (NPLs)

2.1.1 The concept of non-performing loans (NPLs):

In Vietnam, perspectives on bad debt vary significantly, influenced by the country's economic context and the views of different organizations Commonly used terms for problematic debts include "non-performing loans" (NPL), "doubtful debt," and "problem loans." A loan is typically classified as bad when principal and interest payments are overdue by 90 days or more However, there is currently no unified standard for defining NPLs.

The International Monetary Fund (IMF) defines bad debt as loans where payments are overdue by 90 days or more, or where interest payments have been restructured due to repayment concerns Specifically, a loan is classified as non-performing if it is either late for over 90 days or has uncertain repayment prospects This definition is recognized and utilized by various countries, including China.

The European Central Bank (ECB) defines bad debt as a bank loan account that is over 90 days past due, where the borrower has failed to pay either principal or interest, according to the Basel Committee on Banking Supervision (BCBS) criteria Consequently, the ECB's assessment of bad debt is fundamentally quantitative, focusing on the 90-day delinquency threshold.

The International Accounting Standards (IAS) define a key metric for assessing loan repayment capacity, highlighting the importance of completing loans within 90 days or without delinquency To evaluate a customer's ability to repay debt, analysts often estimate future cash flows or classify loan accounts While this methodology is theoretically sound, its practical application remains complex and is not yet widely adopted in the industry.

In Vietnam, according to the Article 3(8) and Article 3(9) of Chapter I in Circular

According to No 11/2021/TT-NHNN from the State Bank of Vietnam, non-performing loans (NPLs) encompass internal bad debts classified into groups 3, 4, and 5 The NPL ratio represents the proportion of bad debts relative to total debts across groups 1 to 5 NPLs are assessed based on two criteria: quantitatively, group 3 includes debts overdue between 90 to 180 days, group 4 consists of debts overdue from 181 to 360 days, and group 5 comprises debts overdue for over 360 days Qualitatively, bad debts are categorized as subprime, doubtful, and those posing a risk of capital loss.

When comparing the definitions of bad debt from the State Bank of Vietnam, IMF, EBC, and IAS, it is evident that, aside from the IAS definition, the first three organizations align on the quantitative criterion of late repayment exceeding 90 days The State Bank of Vietnam's evaluation of bad debt closely mirrors that of the IMF, incorporating an additional qualitative aspect that considers the borrower's capacity to repay within the 90-day timeframe.

Bad debt is primarily defined by two factors: a loan obligation that is 90 days or more overdue and doubts about the borrower's repayment capability To assess this, the bad debt ratio is calculated by dividing the value of loans overdue for over 90 days by the total outstanding loans, in accordance with Vietnamese Accounting Standards (VAS) applicable to commercial banks.

Tracking non-performing loans (NPLs) is essential for banks and the overall economy, as it directly affects bank profitability While banks earn interest from loans, unsuccessful loans lead to "reverse efficiency," risking the loss of both principal and interest, which diminishes capital for future lending By closely monitoring the bad debt ratio, banks can proactively implement strategies to mitigate potential crises Additionally, understanding the evolving bad debt landscape helps banks establish risk protection measures, enhancing their reputation, competitiveness, and trustworthiness in the financial market, which facilitates easier capital mobilization Effective bad debt management also ensures liquidity, underscoring the importance of monitoring bad debt percentages for economic stability, given banks' critical role in capital flow.

As a result, good bad debt management will boost economic development and avoid financial market cash flow interruptions

Laurin et al (2003) highlight the absence of a unified accounting standard for classifying non-performing loans, which is essential for bank managers to assess repayment status and associated risks Effective debt classification is crucial for implementing measures to mitigate potential issues and maintain the credit quality of loan portfolios, even in adverse scenarios.

Table 2.1 Debt classification of some countries in the world

Country Number of loan groups

Mexico 7 Groups are categorized based on country risk, financial risk, industry risk, and payment history Includes groups with redundancy appropriations: A-1 (0.5%), A-2 (0.99%), B-1 (20%), C-1 (20%- 40%), C2 (40%- 60%), D (60-90%) và nhóm E (100%)

Spain 6 Generral provisions and specific provisions

The general proviosions rate is

0.51% the provisions rate for the last 3 groups is respectively 10%, 25% - 100%, 100%

America 5 Do not have specific provisions

Australia 5 There are no specific provisions on provisioning

Italia 5 There are no specific provisions on provisioning

China 5 Generral provisions and specific provisions

The provisions rate for the 5 groups is respectively 1%, 3%, 25%, 75% và 100%

India 5 Generral provisions and specific provisions

Divided into two types including collateral and unsecured The provision rate depends on different groups

The provision ratio for the last 3 groups respectively is 10%, 50% và 100%

Japan 5 The provision ratio for the last 3 groups respectively is 15%, 70% và 100%

Argentina 5 Generral provisions and specific provisions

Tỷ lệ dự phòng cho

5 nhóm lần lượt là 1%, 3%, 12% và 50%

Russia 4 Generral provisions The provision ratio and specific provisions for the last 3 groups respectively is 20%, 50%, 100% The ratio of group

There are 4 groups: risk-free loans, loans with signs of risk, debts with signs of non- recovery, bad debts

In the G-10 countries, the United States and Germany have established a systematic approach to debt classification, with bank managers responsible for adhering to internal rules and procedures The classification of debt is influenced by each country's economic conditions, financial stability, and monetary policies The Institute of International Finance (IIF) has contributed to this discourse by providing analyses and recommendations for categorizing debt In their 2001 work, "The Treatment of Non-Performance Loans in Macroeconomic Statistics," Adriaan M Bloem and Cornelis N Gorter advocated for referencing the IIF's debt analysis, which outlines five distinct groups for effective debt classification.

 Group 1 – Standard: Debt have paid principal and interest timely Moreover, it does not have any problems with debt payment and is willing to conduct committed

Group 2 – Watch: Borrowers in this category often fail to repay their debts in full and tend to procrastinate Consequently, bank managers should exercise increased vigilance and focus on monitoring these debtors more closely than usual.

 Group 3 – Substandard: Debts without payment in full both principal and interest, 90-day debts overdue, reduced value of collateral, or insolvency resulting in the loss

 Group 4 – Doubtful: Debts determined to be insolvent are more than 180 days overdue

 Group 5 – Loss: Non-collectible debts are more than 365 days overdue

Non-performing loans belong to the last 3 groups In Vietnam, according to Decision 493/2005/NHNN and Circular 02/2013/NHNN regulated that classification loans include 5 groups estimated by qualitative and quantitative methods

Table 2.2 Classification of debt groups in Vietnam

Debt group Quantitative method Qualitative method

1 – Standard Debt less than 10 days overdue or due

Can fully recover principal and interest on time

2 – Watch Debts overdue from 10 to

90 days, debts adjusted for the first-time repayment

It is possible to recover the principal and interest in full but there are signs of impairment in the ability to repay the debt

3 – Substandard Overdue debt 91 – 190 days, first renewal debt, interest exemption or reduction

The debt is not able to recover the principal and interest that go due, there is a possibility of losses

4 – Doubtfull Overdue debt 180-360 There is a high probability days, debt restructuring the second repayment term of losses

5 – Loss Debts that are more than

360 days overdue, debts that have structured debt repayment term 2 times but are still overdue or structured debts for the 3rd time or more

No longer able to recover or lose capital

2.1.3 Causes of non-performing loans (NPLs):

Non-performing loans (NPLs) arise from three primary factors, with the borrower being a significant contributor Borrowers with weak financial management skills may face poor business performance, job losses, or an inability to generate sufficient revenue to repay principal and interest to the bank Additionally, ethical issues may arise when borrowers provide fraudulent information during the loan application process or lack honesty regarding their intended use of bank funds, showing little willingness to repay their debts A lax appraisal process before, during, and after the loan can further increase the likelihood of NPLs.

Secondly, NPLs can also occur dut to the bank‘s side There are 3 main reasons:

Credit concentration in banking occurs when a bank allocates a significant portion of its lending to a single customer, a specific group of customers, a particular industry, or a geographical area, often due to loyalty or strategic reasons This practice can lead to increased credit risk, especially when the sectors or regions with high credit concentration experience fluctuations or insolvency, potentially jeopardizing the bank's financial stability.

The credit granting process often lacks strict regulations, creating loopholes that customers can exploit, which contributes to the risk of non-performing loans (NPLs) Additionally, misjudging customers' financial health can lead banks to offer lower loan terms, while competition with rival banks may prompt them to take on excessive risks.

The professional qualifications and ethics of credit officers play a crucial role in the management of non-performing loans (NPLs) for banks Inadequate due diligence and improper assessment of a customer's financial health can lead to inefficient loans, increasing the likelihood of future defaults Additionally, macroeconomic factors significantly influence NPLs; for instance, the economic recession between 2007 and 2010 resulted in higher NPL rates due to inflation and unemployment Research by Marcucci and Quagliariello (2008) indicates that default rates exhibit a cyclical pattern, decreasing during periods of economic growth and rising during recessions Furthermore, Shu (2002) found that a rise in GDP, inflation, and real estate prices correlates with a decrease in NPL ratios, while higher nominal interest rates are associated with an increase in defaults Overall, both the qualifications of credit officers and macroeconomic conditions are vital determinants of NPL ratios and borrowers' repayment capabilities.

Determinants affect non-performing loans (NPLs)

The previous of non-performing loans : According to consolidated document No.22/VBNH- issued on 04/06/2014, NPL ratios are debts classified into group 3

Sub-standard debt, doubtful debt, and debt capable of losing capital are critical classifications in assessing credit risk The World Bank indicates that commercial banks should maintain a bad debt ratio below 5% to ensure healthy credit activities, ideally between 1% and 3% In Vietnam, Circular 23/2020/TT-NHNN mandates that this ratio should be less than 3%, with higher levels leading to state control over banks and restrictions on activities such as share trading and credit expansion The provision ratio for different loan groups varies, with group 1 at 0%, group 2 at 5%, group 3 at 20%, group 4 at 50%, and group 5 at 100% An increasing provision ratio indicates a rising non-performing loan (NPL) rate, negatively impacting banks' asset quality, credit, and liquidity Credit risk provision expenses reflect the potential hazards of loans and forecast bad debt trends, revealing a bank's risk appetite and management efficiency Consequently, bank administrators must enhance related charges to mitigate liquidity risks, which can inadvertently lead to a rise in bad debts.

Bank leverage is a crucial metric for monitoring and assessing the volatility of a bank's bad loans, as it reflects the extent to which banks utilize borrowed assets to finance operations and optimize returns By employing leveraged financial instruments such as bond loans, stocks, and short-term loans, banks aim to enhance efficiency while managing risks A high leverage ratio indicates significant reliance on borrowed funds to generate income, which in turn elevates credit risk and the likelihood of non-performing loans (NPL) Research, including studies by Louzis et al (2012) and Chaibi et al., highlights a positive correlation between the NPL ratio and bank leverage, emphasizing the importance of this relationship in financial assessments.

Research by Muhamad Waquas et al (2015), Pham Duong Phuong Thao and Nguyen Linh Dan (2018), and others indicates a complex relationship between leverage ratios and bank performance Modestos Pappas et al (2018) found that an increase in leverage ratios correlates with reduced volatility in non-performing loans (NPLs), while a decrease leads to higher NPLs This suggests that leverage ratios are crucial indicators of a bank's performance, enabling analysts and shareholders to evaluate asset utilization and assess the risk tolerance of bank managers (Hu et al., 2004; Jimenez and Saurina, 2006; Boudriga et al., 2009; Nikolaidou and Vogiazas, 2011).

The size of a bank is a crucial indicator of its market capacity and competitiveness, with larger banks typically achieving a greater market share in the credit market through wholesale banking services In contrast, smaller banks often focus on retail banking During economic turmoil, the volume of bad debts in wholesale banking tends to rise significantly compared to retail banking Empirical studies, including those by Tran Huy Hoang and Le Thi My Tien (2022), indicate a correlation between bank size and non-performing loans (NPL), revealing that a 1-unit increase in bank size results in a 0.0931 unit increase in bad debts Similar findings have been reported in research by Rajan and Dhal (2003) and Ghosh (2015).

Research indicates a complex relationship between bank size and non-performing loans (NPLs), with some studies showing a positive correlation, while others suggest an inverse relationship Larger banks, as noted by Salas and Suarina (2002), possess substantial total assets, which enable them to enhance credit operations and improve risk management Their significant scale and market share allow for greater diversification in loan activities, effectively distributing interest rate risks and reducing credit concentration Additionally, larger banks have a wider selection of credit consumers compared to their smaller counterparts.

High-return banks are less inclined to engage in high-risk credit activities, as their profitability reduces the pressure to seek risky profits These banks can select clients with strong credit histories, leading to a lower non-performing loan (NPL) ratio in the future Increased earnings decrease the likelihood of bank executives pursuing risky investments, thereby reducing the chances of loans becoming bad debts Conversely, declining profits may prompt managers to take on greater risks, raising the potential for loans to turn into NPLs Moreover, substantial profits equip banks with the capital necessary to manage bad loans and enhance their risk management systems, as supported by various studies (Hu et al., 2014; Messai and Jouini, 2013; Makri, Tsagkanos, and Bellas, 2014; Nguyen Thi Hong Vinh, 2015; Bui Duy Tung and Dang Thi Bach Van, 2015).

2.2.2 Macroeconomic factors of the economy:

Numerous global studies have highlighted the significant impact of macroeconomic conditions on the bad debt ratios of commercial banks Key indicators such as GDP, inflation, and unemployment are frequently analyzed in this context For instance, Nkusu (2011) demonstrates that worsening macroeconomic conditions elevate the risk levels for banks Additionally, research by Festi, Kavkler, and Repina further supports these findings.

(2011) contend that export-boosting economies raise the risk of default and have a direct impact on bank financial stability Alternatively, Fainstein and Novikov

Macroeconomic conditions significantly influence bad debt, particularly the non-performing loan (NPL) ratio of commercial banks in Vietnam Inflation, defined as a continuous rise in the overall price of goods and services, reduces the purchasing power of currency, leading to decreased profits for many businesses and affecting borrowers' repayment abilities Various studies indicate that inflation diminishes the real value of fixed-rate loans, aiding borrowers in repayment Conversely, inflation can also negatively impact consumer revenue and credit solvency Research shows that as GDP grows and consumer goods price inflation rises, the ratio of non-performing loans to credits decreases However, when lending interest rates are variable, banks may adjust rates to maintain real interest rates, potentially increasing the ratio of defaulted loans.

In 2012, GDP and inflation emerged as the primary factors adversely affecting bad debt levels, with inflation potentially influencing the non-performing loan (NPL) ratio in a similar manner.

The World Health Organization (WHO) defines GDP growth as the annual rate of change in a country's gross domestic product, calculated in local currency and based on World Bank estimates from United Nations data According to Pham Xuan Quynh (2017), GDP growth signifies an expanding economy, creating new business opportunities and enabling banks to effectively manage capital flows Research by Do Quynh Anh and Nguyen Duc Hung (2013) highlights that inflation and GDP growth influence non-performing loans in Vietnamese commercial banks, revealing a negative correlation between economic growth and bad debt, a finding supported by various studies, including those by Salas and Suarina (2002) and Jimenez and Saurina (2005).

Research indicates that business cycle outcomes significantly influence the ability to repay interest and principal During periods of economic growth, wages increase, enhancing individuals' and organizations' financial capacities, which in turn lowers the non-performing loan ratios for banks Conversely, during economic downturns, growth rates decline, leading to reduced income from business activities and diminishing repayment abilities This results in a higher percentage of bad loans for banks.

Economic policy uncertainties arise from shifts in government policies, impacting decisions related to investment, consumption, and savings, as measured by the Economic Policy Uncertainty (EPU) index Following the 2008 global financial crisis, the concept of economic policy uncertainty has garnered increased attention from regulators due to its significant effects on microeconomic factors The situation has deteriorated further with the onset of the Covid-19 pandemic, complicating the ability of borrowers to assess credit risk during periods of high EPU This uncertainty leads to a decline in credit activities, ultimately stifling investment and other economic pursuits as capital becomes less mobile within the economy.

The "domino" effect of economic policy uncertainty (EPU) leads to stagnation across various industries, reducing income and affecting customers' debt repayment capabilities A study by Karadima (2020) analyzed the EPU's impact on non-performing loans (NPLs) using data from 500 banks in France, Germany, Italy, and Spain from 2005 to 2017 The research highlighted that selected banks encountered significant challenges following the financial crisis, with EPU identified as a key contributing factor Overall, the findings indicate that EPU negatively influences NPL levels.

A study conducted in Vietnam in 2021 examined the relationship between Economic Policy Uncertainty (EPU) and Non-Performing Loans (NPL) by analyzing data from 900 European banks between 2005 and 2014 using the GMM regression approach The results indicated that an increase in the EPU index negatively affects bank stability, leading to a rise in credit risk, particularly non-performing loans Mohamad Hashem Botshekan and colleagues further explored the impact of global economic policy uncertainty on NPL, utilizing Dynamic Conditional Correlation Generalized Autoregressive Conditional Heteroskedasticity to assess the relationship Their findings revealed that fluctuations in economic policies, such as social isolation and monetary tightening, adversely affect bad debt levels, with high EPU correlating to increased NPL and heightened credit risk for Iranian banks To address the limitations of the EPU index, which only evaluates economic policy insecurity in 29 countries, Ahir et al (2018) developed the World Uncertainty Index (WUI) to measure global economic instability, allowing for more comprehensive research results.

Foundation theory

Asymmetric information, a concept introduced by George Akerlof in 1970, refers to the unequal knowledge between a vendor and a customer regarding a product, service, or investment This phenomenon is particularly significant in the financial sector, especially in credit activities, where borrowers possess better awareness of their financial capabilities, history, and collateral value than lenders Consequently, banks may find themselves at a disadvantage, leading to a scenario where they invest in potentially risky loans while overlooking other viable customers This imbalance can result in an increased rate of bad debt, as highlighted by the research of Greenwald, Stiglitz, and Weiss.

In their 1984 study, Myers and Majluf address the issue of adverse selection in investment decisions, highlighting that effective differentiation between good and bad customers is hindered by information asymmetry during a bank's due diligence process.

Asymmetric information leads to moral hazard issues, particularly in credit activities, as highlighted by Du et al (2005) in their research on Vietnam Poor diagnosis of information problems significantly contributes to high bad debt ratios In cases of moral hazard, borrowers may provide inaccurate or incomplete information regarding their credit history, financial capacity, or legal status to secure larger loans, undermining the bank's due diligence This situation increases the risk of bad loans, which can destabilize banks financially Consequently, banks must allocate more resources for risk provisions, negatively impacting profitability due to higher provisioning costs and reliance on earned profits for funding.

Asymmetric information theory, also known as capital gain order theory, underpins hierarchical order theory, which influences how businesses prioritize funding sources Companies typically favor internal capital, primarily reinvested profits, followed by borrowed funds, and lastly equity This prioritization splits the hierarchical order, as highlighted by Myers (1984), who noted that information asymmetry in financial markets leads to increased costs In the banking sector, the unique nature of capital flow and monetary policy distinguishes banks from other businesses Understanding the categorization of financing sources enables banks to assess risk levels and ensure effective capital allocation in compliance with regulations Hierarchical order theory further aids banks in formulating borrowing strategies and managing risk to maintain profitability.

Adam Smith, a British economist, proposed the principle of economies of scale in

In 1776, the concept of economies of scale emerged, highlighting how larger businesses can achieve significant cost reductions and competitive advantages over smaller firms, ultimately driving economic growth As companies grow, they benefit from worker specialization and integrated technologies that enhance manufacturing efficiency By distributing internal costs across a broader range of products, they can significantly lower unit prices In the financial sector, customer data is crucial for credit activities and capital mobilization; major banks leverage extensive client information to facilitate cross-selling among their branches Additionally, the scale of larger banks fosters brand familiarity, enhancing customer trust and positioning them as more reliable than smaller institutions, thereby encouraging greater acceptance of their credit and capital mobilization services.

Le Ha Diem Chi (2022) explores the correlation between the size and performance of Vietnamese commercial banks, revealing that larger banks tend to generate higher profits than their smaller counterparts due to cost advantages These large banks capture a significant market share in the financial sector, allowing them to selectively choose quality clients for credit activities Additionally, by minimizing information asymmetry, larger banks enhance their credit risk management and human resource capabilities, thereby reducing potential losses and improving their ability to manage future bad debt percentages and resolve outstanding debts from previous years (Salas and Suarina, 2002).

The classical theory of economic development, rooted in the works of economists like Adam Smith, R Malthus, and David Ricardo, provides a foundational understanding of economic growth but tends to overstate the importance of natural resources Following the Great Depression (1929-1933), this theory struggled to explain the economic realities of the time In response, John Maynard Keynes published "The General Theory of Employment, Interest, and Money" in 1936, advocating for active government intervention to regulate and sustain economic growth Keynes emphasized that achieving economic equilibrium, reducing unemployment, and fostering GDP growth require strategic government policies that stimulate demand, investment, consumption, and production He highlighted the significance of monetary policy, suggesting that lowering interest rates could inject more money into the economy and facilitate corporate expansion Furthermore, Keynes proposed "controlled inflation" as a means to enhance company profits, provided manufacturing costs remain stable Efficient business operations lead to GDP growth, increased earnings for individuals and organizations, and improved financial stability, ultimately reducing the bad debt ratio Thus, GDP growth and inflation are interconnected with the bad debt ratio.

Government intervention is essential for market regulation, as national agencies formulate policies that drive economic development and stabilize financial markets The dynamic nature of global economic integration necessitates continuous policy adjustments to address varying circumstances and events, such as the Covid-19 pandemic For instance, Vietnam's WUI index, which was zero in 2018 and 2019, surged to 5.59% in 2020 due to pandemic-related economic strategies and financial support for struggling firms These fluctuations significantly affect banking operations, leading to changes in interest rates and monetary policies, ultimately influencing the bank's non-performing loan (NPL) ratio.

The author concludes that Asymmetric Information Theory highlights the influence of historical bad debts, credit risk provisioning, and profitability on bank bad debt ratios.

Pecking Order Theory highlights that effective risk management and funding prioritization reduce credit system volatility and enhance bank profitability Consequently, factors such as credit risk provisioning, bank leverage, and earnings significantly influence the bad debt ratios of banks.

Thirdly, the author discovered that bank size has an effect on a bank's NPL ratio using the Economy of Scale Theory

After examining John Maynard Keynes' contemporary theory of economic growth, it was found that the Non-Performing Loan (NPL) ratio is influenced by several key factors, including the GDP growth rate, inflation, and the World Uncertainty Index.

RESEARCH MODELS AND METHODS

Research process

The theoretical foundation and pertinent investigations described in the previous chapters contribute to the model's development The research is carried out in the following manner:

Step 1: Summarize previous research from other countries as well as domestic research papers From there, synthesize theories and results of empirical studies to build a theoretical basis suitable to the topic of research

Step 2: Based on the background theory that has been summarized, the author proceeds to build the model At the same time, the author provides a way to measure and hypothesize research for each variable in the model

Research overview Model building Data collection

Step 3: Gather data from annual financial accounts of Vietnamese commercial banks, Fiin Pro, the World Bank, and FRED

Step 4: Compile statistics summarizing the research sample of commercial banks used to present the broadest picture of the study

Step 5: The author conducts correlation testing to examine the relationship between variables in the model

Step 6: The regression model (Pooled OLS, FEM, REM) is estimated by the author

Then, identify the model that best fits the research data

Step 7: Conduct a flaw inspection on the chosen model Variable variation, autocorrelation, and endogenous defects are examples of flaws

Step 8: The author processes the endogenous problem with SGMM regression.

Research models

The author identifies key factors consistently highlighted in various studies that are pertinent to Vietnam's economic landscape during the specified period Subsequently, a study model is developed based on these findings.

𝑵𝑷𝑳 𝒊,𝒕 : Non-performing loan ratio of Bank i in the t year

𝑵𝑷𝑳 𝒊,𝒕−𝟏 : NPL ratio of bank i in the t-1st year

𝑳𝒍𝒓 𝒊,𝒕 : Loan loss reserve ratio of Bank i in the year t

𝑳𝒆𝒗 𝒊,𝒕 : Leverage of Bank i in the year t

𝑺𝒊𝒛𝒆 𝒊,𝒕 : Size of Bank i in the year t

𝑹𝑶𝑨 𝒊,𝒕 : Return on Assets of Bank i in the year t

𝑮𝑫𝑷_𝑮𝑹 𝒕 : GDP Growth Rate in the year t

𝑾𝑼𝑰 𝒕 : World Uncertainty Index in the year t

The Non-Performing Loan (NPL) ratio is a critical metric used to evaluate the proportion of debts that borrowers are unable or unwilling to repay It reflects the number of non-performing loans for every 100 dong lent by banks A bank is deemed to have acceptable credit quality when its NPL ratio is below 3% Conversely, when this ratio surpasses the industry average, it signals potential concerns regarding the company's financial health.

NPL1 LLR LEV SIZE ROA macro variables

The WUI indicates that the management of credit risk is insufficient, as it reveals that debts from customers are not being recovered effectively A lower WUI ratio than the industry average, especially if it shows a declining trend, suggests that banks are excelling in their loan management practices.

𝑵𝐏𝐋 = loan balance group 3 + loan balance group 4 + loan balance group 5

The non-performing loan (NPL) ratio from the previous year serves as a crucial indicator of ineffective loans that remain unrecovered An increase in the current year's NPL ratio can be attributed to these outstanding loans, highlighting the ongoing challenges in loan recovery (Makri, Tsagkanos, and Bellas, 2014).

The Loan Loss Reserve Ratio (LLR) is a crucial metric that reflects the expenses associated with setting aside provisions for credit risks It indicates the proportion of costs that can cover potential bad debts when interest and principal payments are not received This ratio is influenced by the classification of debt groups, and the necessary figures are derived from the balance sheet The computation of the LLR is essential for assessing financial stability and risk management in lending institutions.

Bank leverage (𝑳𝒆𝒗 𝒊,𝒕 ): This is an independent variable measured by debt obligations to a bank's total assets

The size of a bank, indicated by the natural logarithm of its total assets in year t, serves as a key metric for assessing its competitiveness and market share in the capital market.

Return on Assets (ROA) is a key metric for assessing a bank's performance, calculated by dividing net income by total assets This indicator provides valuable insights into how effectively a bank utilizes its assets to generate profit.

Inflation (Infl) is an independent micro-variable that reflects the rate of increase or decrease in the Consumer Price Index (CPI) It is measured using data collected from the General Statistics Office of Vietnam (GSO).

Gross Domestic Product Growth Rate (𝑮𝑫𝑷_𝑮𝑹 𝒕 ): This is the percentage of growth in gross domestic product over the same quarter previous year

The World Uncertainty Index (WUI) measures the ambiguity and unpredictability of economic policies globally Developed by Ahir et al., the index quantifies the frequency of ambiguous phrases found in the Economic Intelligence Unit (EIU) nation reports.

Following a review of relevant research and prior empirical evidence, the author develops hypotheses concerning the relationship of micro and macro factors with NPLs

 Hypothesis 1 (H1): Non-performing loans in the previous year (t-

1) is expected to have a positive with non-performing loans in the current year (t)

The previous year's non-performing loans (NPLs) significantly influence the current year's NPL ratio, highlighting the importance of an effective credit management system to prevent the accumulation of defaulted loans Empirical studies by Do and Nguyen (2014) and Nguyen Thi Hong Vinh (2015) reveal a positive correlation between past and present NPLs, underscoring the need for proactive measures in credit oversight.

 Hypothesis 2 (H2): Loan loss reserve ratio has a positive relationship with the NPLs

As provisioning ratios increase, bank managers face heightened concerns regarding credit risk and unpredictable operating costs, reflecting a similar trend observed in the non-performing loan (NPL) ratio Empirical studies, including those by Berger and DeYoung (1997), the Financial Stability Board (FSB, 2017), Barth et al (2013), Ekanayake and Azeez (2015), and Nguyen Thi Anh Hoa (2021), provide evidence of the interconnected relationship between provisioning ratios and NPL ratios, highlighting a consistent pattern in these financial metrics.

 Hypothesis 3 (H3): Bank leverage has a positive effect on the non-performing loans

Operating leverage is a crucial metric that assesses a bank's utilization of assets and equity A high operating leverage ratio indicates that the bank relies significantly on borrowed assets to generate income, which heightens credit risk and may lead to an increase in bad debt ratios This heightened risk can stem from over-concentration in specific industries or from customers facing substantial risks As these risks escalate, the probability of encountering bad debts also rises Research by Louzis et al (2012), Chaibi et al (2015), Muhamad Waquas et al (2017), and Pham Duong Phuong Thao and Nguyen Linh Dan (2018) supports this interrelationship.

 Hypothesis 4 (H4): Bank size (SIZE) negatively impacts on NPLs

Large-scale banks benefit from established market share in the capital market, allowing them to avoid the pressure of maximizing profits at all costs In contrast, smaller banks must pursue high returns by engaging in riskier projects Additionally, larger banks have greater access to customers with strong financial backgrounds and credit histories, which helps mitigate lending risks This dynamic indicates that bank size inversely correlates with the proportion of bad loans Empirical evidence supporting this hypothesis has been documented in studies by Salas and Suarina (2002), Louzis et al (2010), Hu et al (2014), and Ghosh (2015).

 Hypothesis 5 (H5): Return on Assets (ROA) has the opposite effect on NPLs

The relationship between Return on Assets (ROA) and Non-Performing Loans (NPL) is deeply intertwined, with each variable exerting a significant influence on the other As a key indicator of a bank's profitability and operational effectiveness, ROA has a substantial impact on NPL Notably, previous research has consistently shown an inverse correlation between ROA volatility and NPL, which is understandable given the high costs associated with NPL treatment.

Return on Assets (ROA) is a key metric for evaluating a bank's profitability, with higher ROA indicating greater profitability This strong performance enables banks to strengthen their credit risk management practices, leading to an inverse relationship between ROA and Non-Performing Loans (NPLs) Empirical studies by Messai and Jounini (2013), Nguyen Thi Nhu Quynh, Le Dinh Luan, and Le Thi Huong Mai (2018), as well as Doan Thi Thanh Hang (2020), support these findings.

 Hypothesis 6 (H6): Inflation is positively correlated with NPLs

RESEARCH RESULTS

Statistics describe and consider correlations Linear multi-additive in the

The author utilized Stata 15 software to conduct statistical analyses on a sample of 242 observations across nine independent variables Descriptive statistics were generated, revealing key metrics such as standard deviation, mean, maximum, and minimum values for non-performing loans (NPL1), credit risk provision (LLR), bank leverage (LEV), as well as additional variables including bank size (SIZE), Return on Assets (ROA), inflation (INFL), GDP growth rate (GDP_GE), and the World Uncertainty Index (WUI) The results are presented in Table 4.1.

Table 4.1 Statistics describing variables in the research model

(Source: Author's data processing results from Stata 15 software)

NPL has an average value of 0.0224, ranging from the smallest value from 0.0021 (PVcombank in 2019) to the largest value of 0.0081 (TP Bank in 2015) with a standard deviation of 0.0136

NPL1 has an average value of 0.0230, ranging from the smallest value from 0.0021 (PVcombank in 2019) to the maximum value of 0.0081 (TP Bank in 2015) with a standard deviation of 0.0137

LLR has an average value of 0.0438, which ranges from the smallest value from 0 (AGR in 2012) to the maximum value of 0.9732 (Pvcombank in 2012) with a standard deviation of 0.1340

The LEV has an average value of 0.9096, which ranges from the smallest value from 0.7617 (SGB in 2012) to the maximum value of 0.9740 (SCB in 2021) with a standard deviation of 0.0382

SIZE has an average value of 8.1467, ranging from the smallest value from 7.1233 (Baovietbank in 2012) to the largest value of 9.246 (BID in 2021) with a standard deviation of 0.0207

ROA has an average value of 0.0142, which ranges from the smallest value from 0.000083 (2016 BVB) to the maximum value of 0.1476 (2012 NVB) with a standard deviation of 0.0207

INFL has an average value of 0.0379, which ranges from the smallest value from 0.0063 (in 2015) to the maximum value of 0.0909 (in 2012) with a standard deviation of 0.0229

GDP_GR has an average value of 0.1158, which ranges from the smallest value from 0.026 (in 2021) to the maximum value of 0.6668 (in 2015) with a standard deviation of 0.1846

WUI has a mean value of 0.0906, which ranges from the smallest value from 0.0629 (in 2014) to the maximum value of 0.1438 (in 2019) with a standard deviation of 0.0245

4.1.2 Testing the correlation between variables in the study model

Table 4.2 Correlation matrix between variables in the model

NPL NPL1 LLR LEV SIZE ROA INFL GDP_G

(Source: Author's data processing results from Stata 15 software)

The correlation matrix illustrates the relationship between variables in a model, with values ranging from -1 to 1 A correlation of 1 or -1 indicates a perfect relationship, while a value of 0 signifies no connection In simpler terms, as the correlation value approaches -1 or 1, the likelihood of a relationship increases, whereas values closer to 0 indicate a weaker correlation.

Negative correlation between variable pairs indicates an inverse relationship, where an increase in one variable leads to a decrease in the other Conversely, a positive correlation signifies that both variables move in the same direction, either rising or falling together.

According to the comparison criterion established by Farrar and Glauber (1967), the correlation coefficients among the variables in the matrix table did not exceed the threshold of 0.8, indicating a low likelihood of multilinear multiplication The highest correlation was observed between SIZE and LEV at 0.5995, while the next highest correlation of 0.5303 was found between NPL1 and NPL Conversely, GDP_GR and LEV exhibited the lowest correlation at -0.0020 Overall, the results indicate that the correlation among the model's independent variables is relatively low.

Multicollinearity inspection of the research model

Prior to implementing the Pooled OLS research method, the author will perform a VIF coefficient test to assess linear multi-additivity within the model This analysis will help identify any potential issues, allowing for appropriate adjustments to mitigate standard error discrepancies and ensure the accuracy of the estimated variables.

Table 4.3 Test results for multicollinearity in the model

(Source: Author's data processing results from Stata 15 software)

The analysis of the test results presented in Table 4.3 indicates that the average Variance Inflation Factor (VIF) coefficient is 1.40 Additionally, the VIF values for the independent variables are all below 10, suggesting that the regression model is free from linear multicollinearity issues.

In conclusion, the correlation matrix and VIF testing indicate that there are no significant correlations or linear multi-collinearity issues among the independent variables This finding confirms that the model is appropriate for proceeding with linear regression implementation.

Results of regression model estimation

4.4.1 OLS, FEM and REM synthetic regression model

Table 4.4 Regression model results according to OLS, FEM and REM

(Source: Author's data processing results from Stata 15 software)

(Note: *, **, *** indicate statistical significance at the 1%, 5% and 10% levels, respectively)

The OLS model results indicate that out of eight independent variables, three are statistically significant Notably, the variable NPL1 exhibits the strongest impact on the dependent variable NPL, with a coefficient of 0.4607 Both NPL1 and INFL are significant at the 1% level, while SIZE is significant at the 10% level Conversely, the variables LLR, LEV, ROA, GDP_GR, and WUI show no statistical significance, suggesting they do not influence NPL.

The regression analysis using the FEM approach identified three statistically significant variables, with NPL1 and INFL showing significance at the 1% level Additionally, ROA demonstrated significance at the 10% level, with a value of 0.1557 However, the study found no evidence that LLR, LEV, SIZE, GDP, or WUI have any influence on NPL.

The analysis in Table 4.4 reveals that out of eight independent variables, three exhibit statistical significance Specifically, NPL1 and INFL are significant at the 1% level, while SIZE shows significance at the 10% level with a value of -0.0027 Conversely, the variables LLR, LEV, ROA, GDP, and WUI do not demonstrate statistical significance within the model.

A comparison of the 4.4 regression models indicates that each model identifies three statistically significant variables out of eight Notably, NPL1 consistently shows the strongest influence across all models, with coefficient values of 0.4607, 0.2663, and 0.0407 Both NPL1 and INFL demonstrate a significant relationship at the 1% level, while the third significant variable varies slightly among the models In the OLS and REM models, SIZE is significant at the 10% level, a trend not observed in the FEM model, which instead highlights ROA as significant at the same level Conversely, LEV, GDP, and WUI show no statistical significance in any of the models analyzed.

The author evaluates the appropriate research method for the model by performing statistical tests, including the F-test to compare Ordinary Least Squares (OLS) and Fixed Effects Model (FEM), as well as the Hausman test to differentiate between Fixed Effects Model (FEM) and Random Effects Model (REM).

Results of testing and selecting suitable models

 The result of F-test (Fisher test)

F-test is also known as pooling test This test is used to choose between 2 Pooled OLS and FEM models with two hypotheses as follows:

H0: There is no specific impact between each object (Pooled OLS is most suitable)

H1: There is a specific impact between each audience (FEM is most appropriate)

Table 4.5 The result of Fisher test

(Source: Author's data processing results from Stata 15 software)

At a significance level of 5%, the Prob value of 0.0189 indicates that the null hypothesis (H0), which asserts that each variable has a unique existence, is rejected This finding suggests that Ordinary Least Squares (OLS) is not the most suitable model for the analysis Consequently, the author identifies the Fixed Effects Model (FEM) as the preferred modeling approach.

Based on the F-test results, the author decided that FEM was a better fit than OLS

As a result, the author proceeds to use the Hausman test method to compare two models, FEM and REM, based on the following hypothesis:

Table 4.6 The result of Hausman accreditation

Test: 𝐻 0 : Differences in coeficients not systematic

(Source: Author's data processing results from Stata 15 software)

The results in Table 4.6 indicate a Prob>chi2 value of 0.0000, which is less than the significance level of 0.05 This leads to the rejection of the null hypothesis (H0) and acceptance of the alternative hypothesis (H1) Consequently, the author concluded that the Finite Element Method (FEM) model is the most appropriate choice.

Following both testing, the author determined that the FEM model was the most appropriate solution for the study Following that, the author inspects the defect in the selected model.

Inspection of selected model defects

4.6.1 The result of testing the autocorellation

The author conducts a Wooldrige test with the following two hypotheses to investigate the autocorrelation:

𝐻 0 : The model has not an autocorrelation

𝐻 1 : The model has an autocorrelation

Table 4.7 The result of testing autocorrelation

Wooldrige test for autocorrelation in panel data

(Source: Author's data processing results from Stata 15 software)

From the table 4.7 results, it shows that Prob = 0.0000 < ∝ = 0.05 with a significance level of 5% This means rejecting the H0 hypothesis, accepting the H1 hypothesis that the model has an autocorrelation phenomenon defect

4.6.2 The result of testing the heteroscedasticity

The author implements a Wald test with two hypotheses to evaluate the phenomena of variable variance:

𝐻 0 : The model has not heteroscedasticity phenomena

𝐻 1 : The model has heteroscedasticity phenomena

Table 4.8 The result of testing heteroscedasticity

Wald test: H0: sigma(i)^2 = sigma^2 for all i

(Source: Author's data processing results from Stata 15 software)

According to the table results, Prob = 0.0000 < 0.05 with a significance level of 5% This entails rejecting the H0 hypothesis and adopting the H1 hypothesis of variable variance in the model

The author discovered a pattern of both faults, which is the phenomenon of self- correlation and the phenomenon of variable variance, after completing defect tests

To address the issue of non-performing loans (NPLs) impacting bank profitability, the author employs FGLS regression analysis The study highlights the NPL1 variable, indicating that bad debts from the previous period directly affect banks' financial performance due to revenue and expenditure imbalances Nguyen Thi Branch (2009) notes that ineffective loan grants lead to a lack of interest and principal income, forcing banks to seek external capital for operational expenses and interest payments This necessity, coupled with the need to set aside provisions for future loans, increases operational costs, diminishes financial capacity, and ultimately reduces profitability Aremu et al (2013) further confirm that past non-performing loans significantly affect bank profitability both in the short and long term, a finding supported by Muhamad Bilal et al (2013), which emphasizes the prior period's NPLs as a crucial factor in determining commercial bank profitability.

Overcoming model defects using SGMM method

After conducting defect tests, the author identified that the model exhibits both autocorrelation and heteroscedasticity phenomena However, the presence of endogenous factors is suspected, particularly due to the inclusion of a lagged dependent variable as an independent variable, which suggests a dynamic panel data model that may contain endogenous variables, as noted by Richard Blundell and Stephen Bond (1998) The study also revealed a bidirectional relationship between prior non-performing loans (NPL1) and bank profitability (ROA) To address the issues of autocorrelation, heteroscedasticity, and endogeneity, the author recommends employing the System Generalized Method of Moments (SGMM), as suggested by Richard Blundell (1998), Nguyen Thi Hong Vinh (2015), and Pham Duong Phuong Thao & Nguyen Linh Dan (2018) Consequently, the author selected instrumental variables including 𝑳𝒍𝒓 𝒊,𝒕−𝟏, 𝑺𝒊𝒛𝒆 𝒊,𝒕−𝟐, 𝑮𝑫𝑷_𝑮𝑹 𝒕−𝟐, and 𝑾𝑼𝑰 𝒕−𝟒.

𝑰𝑵𝑭𝑳 𝒕−𝟐 to fix the defects of the model by System – GMM method as follows:

Table 4.9 The SGMM model estimation result

(Source: Author's data processing results from Stata 15 software)

(Note: *, **, *** indicate statistical significance at the 1%, 5% and 10% levels, respectively)

SGMM regression results show that 6 variables out of 8 are statistically significant

At the 1% significance level, the analysis identifies three key variables influencing the dependent variable NPL: LLR, SIZE, and NPL1, with LLR demonstrating the strongest effect at a coefficient of 0.5650 At the 5% significance level, the WUI variable is present with a coefficient of 0.1415 Additionally, ROA and GDP_GR are significant at the 10% level, showing coefficient values of -0.1476 and 0.1114, respectively However, there is no evidence to suggest that LEV and INFL affect the NPL variable.

Table 4.9 indicates that the number of instruments (24) is less than the number of groups (28), confirming that the Sargan test (1985) in GMM model estimation is robust Additionally, the P-value from Arellano and Bond's (1991) AR(2) test is 0.780, which is significantly higher than the threshold of 0.780.

The GMM estimate shows no correlation phenomenon, with a value of 0.780 exceeding 0.1 The Sargan test, which evaluates the appropriateness of instrumental variables in the GMM model, indicates a significance level of P > 10% > 5% (0.685 > 0.1 > 0.05) This suggests that the model is free from endogenous variables when exceeding 10% and demonstrates no self-correlation and error when above 5% Additionally, the Sargan test coefficient greater than 0.05 confirms the reasonableness of the instrumental variables used Consequently, the GMM model meets the necessary conditions for analysis, statistical inference, and serves as the final result of the study.

In summary, the research model has the following equation:

The analysis reveals that independent variables such as NPL i,t−1, LLR i,t, WUI t, and GDP_GR t positively influence the dependent variable NPL, with statistical significance at 1%, 5%, and 10% Notably, LLR and NPL1 exhibit the strongest positive effects on NPL Conversely, the ROA variable demonstrates the most significant negative impact on NPL, with a statistical significance level of 10%, while the SIZE variable also negatively affects NPL, achieving significance at 1% Additionally, the variables LEV and INFL were found to be statistically insignificant, as their p-values exceeded 10%.

Table 4.10 Compare research results with hypotheses

Variable Hypothesis Result Statistical significance

CONCLUSION AND RECOMMENDATIONS

Conclusion

By examining data from 28 joint-stock commercial banks in Vietnam from 2012 to

In 2022, an analysis of the impact of macroeconomic conditions and bank characteristics on commercial banks' non-performing loan ratios revealed several influential factors The study found that previous non-performing loans, risk provision ratios, GDP growth rates, and the WUI positively correlated with bad debt ratios, while bank size and return on assets (ROA) had a negative impact Notably, bank leverage and inflation were not statistically significant in the model, diverging from initial estimates.

The study addresses the bad debt situation of Vietnamese commercial banks from 2012 to 2021, highlighting its urgency and key influencing factors Chapter 4 reveals the extent of these impacts and proposes tailored solutions to mitigate bad debts for each identified factor Additionally, the research draws on global best practices in managing bad debts, offering relevant recommendations for Vietnam's economic context aimed at commercial bank managers and the State Bank.

Solution proposal

5.2.1 The provision for credit risk:

The provision for credit risk serves as an indicator of future trends in bad loan ratios, with an increase suggesting heightened risk in variable loans To mitigate credit risk, banks should focus on sustainable credit development by increasing the proportion of loans backed by production assets It is essential for banks to actively pursue recovery options for outstanding debts that have utilized risk provisions Additionally, diversifying investment portfolios is crucial to prevent credit concentration, which can exacerbate credit risk.

The study indicates that larger banks tend to have a lower non-performing loan (NPL) ratio However, to enhance their overall asset base and size, banks need to implement effective restructuring strategies It is crucial for banks to manage the rapid growth of immediate credit, as this can lead to significant credit risks and increased bad debts.

This study indicates that improving return on total assets (ROA) can effectively decrease non-performing loans, suggesting that banks should focus on enhancing their ROA to mitigate bad debt levels During periods of successful expansion, it is crucial for banks to prioritize portfolio diversification and implement strict risk management practices Additionally, to lower transaction costs, banks should enhance the quality of their products and services, while also diversifying non-credit offerings that can be securely exchanged in a robust electronic network environment.

5.2.4 GDP growth rate and World Uncertainty Index (GDP_GR and WUI)

GDP and WUI growth rates are key macroeconomic factors that commercial banks cannot directly influence Nevertheless, bank managers have the ability to forecast these indicators and develop strategic plans to navigate global economic fluctuations effectively.

Recommendations

Based on an analysis of global debt handling policies, the author presents several tailored proposals that align with the unique economic conditions and context of Vietnam.

To enhance the effectiveness of risk management systems, it is crucial for banks to address the strong correlation between the non-performing loan (NPL) ratio from previous periods and current bad debt levels Research indicates that inadequate provisioning for credit risks is linked to rising bad debts, highlighting weaknesses in governance and oversight within banks Consequently, many banks fail to allocate sufficient provisions in line with incurred bad debts, leading to a backlog of unresolved old debts while new ones emerge To combat this, banks should strengthen their risk management practices by adopting Basel II international standards, which only 18 commercial banks had implemented by the end of 2020 This framework enables more thorough assessments of clients' repayment capabilities, thereby mitigating the risks associated with lending to potentially unreliable borrowers Additionally, findings suggest that declining profitability correlates with an increase in bad loans, emphasizing the need for commercial banks to prioritize risk management strategies Learning from the experiences of countries like Hungary, China, and Japan, Vietnamese banks can tighten lending conditions and enhance cross-monitoring among senior leaders across departments, thus addressing core factors contributing to credit risks.

To enhance debt collection and improve the non-performing loan (NPL) ratio, commercial banks should actively encourage debt settlements and manage bad debts effectively This includes selling debts to the Vietnam Asset Management Company (VAMC) and utilizing collateral to recover loans Additionally, banks must implement timely support measures, such as securitizing principal and interest loans, to prevent the bankruptcy of viable firms with strong management This approach not only aids struggling businesses but also helps banks secure loan capital.

To achieve scale expansion, Vietnamese commercial banks need a well-balanced restructuring strategy To enhance profitability, banks should refrain from aggressive upsizing through the excessive use of payday loans or by lowering credit approval standards, as these practices could lead to increased credit risk.

To enhance earnings and optimize capital utilization, commercial banks must collaborate closely with risk management Effective leverage can significantly increase operating profits; however, weak governance can lead to a cascading impact on various operations, resulting in higher costs, decreased leverage efficiency, and an elevated bad debt ratio.

Macroeconomic factors significantly influence the bad debts of commercial banks, although these banks cannot directly alter these macro variables To safeguard their assets, banks should anticipate economic events and develop effective coping strategies Incorporating macro variables into stress test algorithms is essential for bank management, as it allows them to create strategies that mitigate the impact of economic shocks and forecast appropriate risk provisions.

To effectively address bad debts, it is crucial to establish a comprehensive legislative framework Currently, the existing regulatory structure only permits two methods for debt acquisition: direct negotiations and auctions, which limits the establishment of a loan value foundation and hampers information sharing Additionally, real estate is the most common collateral for defaulted loans; however, the lack of a functioning real estate market, due to state control and protracted legal procedures, complicates matters Therefore, the State Bank must develop a robust system for loan valuation to enhance the management of bad debts.

Macroeconomic factors, including GDP growth and inflation, play a crucial role in economic stability The State Bank should enhance economic demand and provide financial support to businesses in need of funding Additionally, it is essential to implement prudent monetary policies consistently to control inflation and mitigate the impacts of global economic fluctuations.

The State Bank must enhance its oversight of credit granting activities in high-risk sectors such as real estate, securities, and corporate bond investments Ongoing monitoring and stringent inspections are essential to identify potential risks and legal violations by banking officials and employees By implementing preventive measures and addressing issues promptly, the State Bank can help ensure the overall safety and stability of the banking system.

Bình Nam 2017, Tín dụng tăng trưởng “nóng” có đáng ngại?, truy cập tại < https://cafef.vn/tin-dung-tang-truong-nong-co-dang ngai20170509161411729.chn> [ngày truy cập: 28/08/2023]

Nghiên cứu của Do, A Q và Nguyen, H D (2013) phân tích các yếu tố quyết định đến nợ xấu tại các ngân hàng thương mại Việt Nam Bài viết được trình bày tại hội thảo khoa học về nghiên cứu kinh tế và chính sách, nhấn mạnh tầm quan trọng của việc hiểu rõ các nguyên nhân gây ra nợ xấu để cải thiện tình hình tài chính của các ngân hàng Kết quả nghiên cứu cung cấp cái nhìn sâu sắc về thực trạng nợ xấu, từ đó đề xuất các giải pháp nhằm quản lý rủi ro hiệu quả hơn trong lĩnh vực ngân hàng.

Nguyễn Thị Phương Thúy (2017) đã trình bày kinh nghiệm xử lý nợ xấu của một số quốc gia trong bài viết trên Cổng thông tin điện tử Viện Chiến lược và chính sách tài chính Bài viết cung cấp cái nhìn sâu sắc về các biện pháp hiệu quả trong quản lý nợ xấu, giúp các nhà hoạch định chính sách tham khảo và áp dụng vào thực tiễn Để tìm hiểu thêm, bạn có thể truy cập tại [đây](https://mof.gov.vn/webcenter/portal/vclvcstc/pages_r/l/chi- tiettin?dDocName=MOFUCM108231) (ngày truy cập: 28/08/2023).

Quỳnh, N T N 2018, ‗Các nhân tố tác động đến nợ xấu tại các ngân hàng thương mại Việt Nam‘, Tạp chí khoa học Đại học mở Thành phố Hồ Chí Minh -

Trong bài viết "Bàn về hướng xử lý nợ xấu của hệ thống ngân hàng thương mại Việt Nam" của Đào Thị Hồ Hương, đăng trong Kinh tế và Quản trị kinh doanh, tập 13, số 3, trang 261-274 năm 2015, tác giả phân tích các phương pháp và giải pháp nhằm xử lý nợ xấu trong hệ thống ngân hàng thương mại Việt Nam Bài viết cung cấp cái nhìn sâu sắc về tình hình nợ xấu hiện tại và đề xuất các biện pháp cải thiện hiệu quả quản lý nợ xấu.

I'm sorry, but I can't access external links or content directly However, if you provide the text or key points from the article, I can help you rewrite it while ensuring it remains coherent and SEO-friendly.

Phong, T T., Bằng, T V., & Phương, N S 2015, Các nhân tố ảnh hưởng đến nợ xấu tại các ngân hàng thương mại việt nam

Thảo, P D P., & Đan, N L 2018, ‗Các yếu tố ảnh hưởng đến tỷ lệ nợ xấu của các ngân hàng thương mại cổ phần Việt Nam‘, Tạp chí Chính Sách & Thị trường

Tài chính-Tiền tệ, số 194, trang 1-10

Thông tư 11/2021/TT-NHNN quy định về việc phân loại tài sản có và mức trích lập dự phòng rủi ro trong hoạt động của tổ chức tín dụng và chi nhánh ngân hàng nước ngoài Thông tư này cũng nêu rõ phương pháp trích lập dự phòng và cách sử dụng dự phòng để xử lý các rủi ro phát sinh.

Tô Ngọc Hưng (2012) đã nghiên cứu kinh nghiệm xử lý nợ xấu từ một số quốc gia, từ đó rút ra những bài học quý giá cho Việt Nam Bài viết nhấn mạnh tầm quan trọng của việc áp dụng các biện pháp hiệu quả để quản lý nợ xấu, nhằm đảm bảo sự ổn định của hệ thống ngân hàng và nền kinh tế Thông tin chi tiết có thể được tìm thấy tại Ngân hàng Nhà nước Việt Nam, qua đường link: [ngày truy cập: 28/08/2023]

Tung, B D 2016, ‗Effects of internal factors on onperforming loans of Vietnam‘s commercial banks‘, Journal of Economic Development

LIST OF COMMERCIAL BANKS USED IN MODEL

ABB An Binh Commercial Joint Stock Bank

ACB Asia Commercial Joint Stock Bank

AGRI Vietnam Bank for Agriculture and Rural Development

Baovietbank Bao Viet Joint Stock Commercial Bank

BID Joint Stock Commercial Bank for Investment and Development of

Vietnam BVB Viet Capital Commercial Joint Stock Bank

CTG Vietnam Joint Stock Commercial Bank of Industry and Trade

EIB Vietnam Export Import Commercial Joint Stock Bank

HDB Ho Chi Minh City Development Joint Stock Commercial Bank

KLB Kien Long Commercial Joint Stock Bank

MBB Military Commercial Joint Stock Bank

MSB Vietnam Maritime Commercial Join Stock Bank

NAB Nam A Commercial Joint Stock Bank

NVB National Citizen Commercial Joint Stock Bank

OCB Orient Commercial Joint Stock Bank

PGB Petrolimex Group Commercial Joint Stock Bank

PVcombank Vietnam Public Joint Stock Commercial Bank

SCB Sai Gon Joint Stock Commercial Bank

SGB Saigon Bank for Industry and Trade

SHB Saigon-Hanoi Commercial Joint Stock Bank

SSB Southeast Asia Commercial Joint Stock Bank

STB Saigon Thuong Tin Commercial Joint Stock Bank

TCB Vietnam Technological and Commercial Joint Stock Bank

TPB TienPhong Commercial Joint Stock Bank

VAB Vietnam Asia Commercial Joint Stock Bank

VCB Joint Stock Commercial Bank for Foreign Trade of Vietnam VIB Vietnam International Commercial Joint Stock Bank

VPB Vietnam Prosperity Joint Stock Commercial Bank

RESULTS OF MODEL FROM STATA 13

1 Descriptive statistics result by Stata 15

2 Correlation matrix result by Stata 15

9 The result of autocorellation test

10 The result of S-GMM regression model

THE METHOD TO SOLVE THE NPLs OF COUNTRIES

COUNTRIES AROUND THE WORLD Refering to the NPLs settlement policies of countries around the world:

The bursting of Japan's price bubble in 1991 caused significant turmoil in the banking system, primarily due to the excessive lending to specific industries like real estate and securities, coupled with lax supervision during an inflationary period By the late 1990s, the country faced recession and deflation, resulting in a surge of bad debt Acknowledging that resolving bad debt is essential for economic recovery, the Japanese government initiated reforms focused on this critical issue.

To establish a highly regulated financial system that safeguards the interests of individuals and organizations, it is essential to ensure the seamless operation of payment and settlement services The Financial Service Authority (FSA) aims to enhance its regulatory role by creating a "task force on financial affairs" dedicated to monitoring bad debt settlements in commercial banks Additionally, the government and central bank will strengthen support systems for credit institutions facing liquidity constraints, ensuring they can continue lending activities and effectively manage bad debts.

Banks benefit from government and central bank-established public funds for "special assistance." To ensure the success of the non-performing loan (NPL) settlement program and financial system reform, the Financial Services Agency (FSA) is considering the establishment of a public fund This fund would provide financial support to the banking system in critical situations, allowing access to government resources when necessary.

The Financial Services Agency (FSA) has introduced a new corporate restructuring framework that enhances loan elimination through strategic loan sales and conducts self-assessments of market value when acquiring loans from banks Additionally, the FSA evaluates the provisioning factors for these loans and strengthens company revitalization efforts by providing financial assistance to rebuild firms, in collaboration with major Japanese banks By creating a supportive environment for business recovery, the FSA offers operational support to enterprises with poor credit histories while continuing to finance efficient businesses with viable objectives, recognizing the crucial role of companies in mitigating non-performing loans (NPLs).

The Financial Services Authority (FSA) has introduced a new framework for financial system management that employs a discounted cash flow technique to analyze reserve funds and assess the risk provisioning of asset time periods, while identifying key debtors within the banking system Additionally, the FSA conducts thorough inspections to ensure compliance with asset classification, provisioning, and collateral valuation regulations To enhance transparency and accountability, the FSA will publish findings that highlight discrepancies between its inspection results and those reported by credit institutions, aiming to guide these institutions in aligning with updated standards.

Non-performing loans (NPLs) in China stem from a centrally planned economic system, where state-owned banks provide loans to inefficient state enterprises without thorough credit evaluations, leading to significant credit risk The resolution of bad debts is intertwined with the shift to a market economy and the restructuring of state-owned enterprises and the financial sector To address this issue, the government established four Asset Management Corporations (AMCs), each aligned with a major state bank, to manage bad debts separately from banking operations, enhancing oversight by regulatory bodies The AMCs actively engage in selling, auctioning, and restructuring bad debts, including successful auctions that generated substantial revenue, such as a recent auction to international organizations for 13 billion yuan Additionally, the government has implemented securitization of bad debts, creating various securities to attract investors based on their risk preferences, thereby mitigating the bad debt crisis.

Hungary has effectively reduced its bad loans from 30% to 5% through strategic reform initiatives Key measures included allowing banks to convert problematic loans into 20-year bonds and the establishment of the Bad Debt Recovery Agency, which swaps government bonds for significant bad debts and manages the sale or restructuring of these debts While banks settle residual debts through contracts with the Ministry of Finance, this approach only temporarily mitigates bad debt issues without addressing underlying causes To tackle this, the government intervened by writing off debts for essential state-owned enterprises in exchange for government bonds and recapitalizing banks to achieve an 8% Capital Adequacy Ratio (CAR) This recapitalization involved using government bonds to purchase newly issued shares, thereby increasing state ownership in banks Additionally, the State representative of the bank will oversee the management and supervision of bad debt settlements.