credit risk liquidity risk and bank stability empirical evidence from joint stock commercial banks in vietnam

Concurrently, bank liquidity, reflected by the ratio of cash, deposits at the State Bank of Vietnam SBV and credit institutions to total assets, above 5.80% eliminates the negative effec

RESEARCH INTRODUCTION

REASONS FOR THE RESEARCH

Nowadays, the banking system plays the central role in any country’s financial system (Nguyen Tran Phuc, 2022) Especially in developing nations like Vietnam, the domestic financial system remains at a low level of development and is a typical bank-based system, thus further highlighting the pivotal role of banks as the key driver of financial system development (Le Dinh Hac et al., 2017) This underscores the urgent need to ensure the banking sector stability to effectively promote financial system’s development, hence achieving the goal of economic growth as bank stability is an important factor in gross domestic product (GDP) growth (Jokipii & Monnin, 2013)

Currently, the banking system in Vietnam is experiencing rapid development with a wide range of products and services Concurrently, it is involved deeply in every sector of the economy, contributing significantly to the domestic financial system's growth However, being the major force for the national financial system's development has exerted considerable pressure on Vietnamese banks, especially joint stock commercial banks (JSCBs) These banks both operate for profit, in accordance with legal regulations, and reinforce their role as a strong arm of the government in implementing economic development policies To achieve these goals, Vietnam’s JSCBs primarily conduct credit extensions, a core business activity that mainly contributes to banks’ profit, reflecting their function of financial intermediation in the economy Specifically, Vietnam’s JSCBs transfer funds from surplus entities to deficit entities in the economy through credit provision This process highlights two typical risks faced by banks: credit risk (the borrower's inability to repay loans upon maturity) and liquidity risk (the unexpected withdrawal of funds by depositors), consequently threatening bank stability, as these risks are the main causes of bank failures and bankruptcies (Ahmad et al., 2019) Relevant events adversely affecting the system of Vietnam’s JSCBs has been witnessed throughout recent times, prompting urgent intervention by the State Bank of Vietnam (SBV) In 2015, SBV acquired CBBank, OceanBank, and GPBank at 0 dong to restructure them due to their excessive credit risks from bad debts (Le Thanh, 2023) In addition, SBV placed SCB under special control from October 2022 to address severe liquidity shortages caused by mass deposit withdrawals amid negative information about its weak financial health (Le Minh Son & Le Phan Quynh Trang, 2022) Accordingly, these events, despite occurring at individual banks, can severely affect the stability of the entire system due to the close connection between banks in the system, consequently hindering the development of the domestic financial system and ultimately impeding national economic growth

Therefore, to foster sustainable economic development in Vietnam, it is imperative to ensure stability and sustainability in the domestic financial system, with particular emphasis on its JSCBs However, the presence of credit and liquidity risks poses threats to bank stability Various reviewed studies highlighted that credit and liquidity risks are key determinants of bank stability (Hakimi et al., 2017; Jabra, 2020; Mishi & Khumalo, 2019; Nguyen Duc Trung et al., 2022; Nguyen Minh Ha & Nguyen Ba Huong, 2016; Rupeika-Apoga et al., 2018; Yitayaw et al., 2023) These findings provide a basis for more detailed research into the individual effects of these risks on bank stability For credit risk, studies by Nguyen Quoc Anh & Duong Nguyen Thanh Phuong (2021) and Khan & Yilmaz (2022) find its negative effect, whereas Kửhler (2015) and Dao Le Kieu Oanh et al (2021) confirm its positive impact on bank stability For liquidity risk, Phan et al (2019) and Nguyen Duc Trung et al (2022) highlight its negative influence, while Setiawan et al (2019) observe its positive effect on bank stability Furthermore, given that credit and liquidity risks often co-occur in bank credit activities, there has been a growing research focus on their interactive effect on bank stability (Amara & Mabrouki, 2019; Chai et al., 2022; Ghenimi et al., 2017; Imbierowicz & Rauch, 2014; Lachaab, 2023; Nguyen Kim Chi et al., 2023; Vo Thi Thuy Kieu et al., 2021)

In summary, discussing the effects of credit and liquidity risks on bank stability, the literature review process outlined above reveals that (i) studies primarily utilize data samples from banks in countries other than Vietnam; (ii) even when data samples are collected from Vietnam, research efforts predominantly concentrate on separate effects (rather than interactive or both separate and interactive effects) of these risks on bank stability Therefore, the research "Credit risk, liquidity risk and bank stability: Empirical evidence from Joint Stock Commercial Banks in Vietnam" with its focus on both individual and interactive effects of credit and liquidity risks on the stability of Vietnam’s JSCBs is necessary to address general research gaps identified in prior research Concurrently, the research aims to provide further empirical evidence for this research strand, thus offering recommendations to reinforce the stability of Vietnam’s JSCBs.

RESEARCH OBJECTIVES

The research is aimed at examining the effects of credit risk and liquidity risk on the stability of Vietnam’s JSCBs The results obtained from the research provide an important basis for making recommendations related to the effects of credit risk and liquidity risk to ensure the stability of Vietnam’s JSCBs

To achieve the general objectives stated above, the research focuses on investigating both the individual and interactive effects of credit risk and liquidity risk on the stability of Vietnam’s JSCBs in order to propose relevant recommendations Below are three specific objectives of the research

Firstly, examine the individual effects of credit risk and liquidity risk on the stability of Vietnam’s JSCBs

Secondly, examine the interactive effect of credit risk and liquidity risk on the stability of Vietnam’s JSCBs

Thirdly, propose recommendations related to the effects of credit risk and liquidity risk to ensure the stability of Vietnam’s JSCBs.

RESEARCH QUESTIONS

The research poses three questions aligned with the aforementioned research objectives below

Firstly, how do credit risk and liquidity risk individually affect the stability of

Secondly, how do credit risk and liquidity risk interactively affect the stability of Vietnam’s JSCBs?

Thirdly, what recommendations related to the effects of credit risk and liquidity risk are made to ensure the stability of Vietnam’s JSCBs?

RESEARCH SUBJECTS AND SCOPE

The research is centered on key subjects, including the individual and interactive effects of credit risk and liquidity risk on the stability of Vietnam’s JSCBs

The research is confined within the scope of space and time outlined below

About spatial scope, the research is conducted within 26 JSCBs in Vietnam

(outlined in Appendix 1) to ensure its applicability, data availability and representativeness below

As for applicability of the research, the research focuses on the banking context in Vietnam, a typical country with a predominantly bank-based financial system (Nguyen Tran Phuc, 2022), thereby underscoring the important role of the banking system and its stability in a developing country like Vietnam This contextual choice ensures the applicability of the research and provides it with appropriate recommendations to ensure the stability of Vietnam’s JSCBs

As for availability and representativeness of research data, the research limits data to Vietnam’s JSCBs, excluding other bank types Data is collected from banks listed on Ho Chi Minh Stock Exchange (HOSE), Ha Noi Stock Exchange (HNX) and Unlisted Public Company Market (UPCOM) to ensure availability, accuracy, and reliability of data The research focuses on 26 out of 31 Vietnam’s JSCBs by March 31, 2024 (State Bank of Vietnam, 2024), hence ensuring the representativeness of data for Vietnam’s JSCBs with a representation rate of about 84%

About temporal scope, the research is confined to the period 2013 – 2023 to ensure its thoroughness and applicability This period captured pivotal fluctuations in Vietnam's banking system influenced by significant legal decisions such as Decision 254/QD-TTg and Decision 843/QD-TTg from the Prime Minister to restructure domestic credit institutions and handle their bad debts, Circular 39/2016/TT-NHNN on lending transactions of banks, Circular 41/2016/TT-NHNN on capital adequacy ratio of banks and Circular 11/2021/TT-NHNN on methods of setting up risk provisions of banks from the SBV Focusing on bank credit and liquidity, these legal documents ensure the research’s thoroughness while considering the legal context In addition, the research extends the time frame up to 2023 to ensure its applicability to contemporary banking practices.

RESEARCH METHODOLOGY

To explore the effects of credit risk and liquidity risk on stability of Vietnam’s JSCBs as outlined in the research questions in Section 1.3, the quantitative research methodology is employed Accordingly, the research uses a panel dataset of 26 Vietnam’s JSCBs from 2013 to 2023 to perform essential calculations to create variables for the research model Descriptive statistics are then conducted to provide an overview of model variables, and correlations among model independent variables are analyzed using the correlation matrix and Variance Inflation Factor (VIF) to identify multicollinearity between these variables, determining which ones are suitable to be included in the model Next, the research model is estimated using methods of Pooled Ordinary Least Squares (Pooled OLS), Fixed Effects Model (FEM) and Random Effects Model (REM), and the optimal estimator is selected through F-test and Hausman test The model with chosen estimator is then tested for its defects of autocorrelation and heteroskedasticity through Wooldridge and Breusch-Pagan tests If these defects, along with endogeneity detected by Durbin and Wu-Hausman tests, are identified, the research will apply the method of Generalized Method of Moments (GMM), specifically the two-step system GMM, to address them This is followed by Wald, Hansen, and Arellano-Bond (AR) tests to ensure the robustness and reliability of the estimation results These quantitative methods are implemented using Stata 17 to examine the effects of credit and liquidity risks on the stability of Vietnam’s JSCBs, in response to research questions 1 and 2

In addition, the research employs listing, analysis, comparison, and synthesis to comprehensively review the theoretical framework of credit risk, liquidity risk, and bank stability, along with theories on the risk-bank stability relationships Furthermore, empirical evidence related to the effects of these risks on bank stability is explored using the aforementioned methods to identify research gaps to determine an appropriate research orientation that aligns with the operational realities and data accessibility of Vietnam’s JSCBs, thereby partly addressing research questions 1 and

2 Additionally, the research continues to use analysis and synthesis to discuss the quantitative research results, in response to research question 3 about recommendations to ensure the stability of Vietnam’s JSCBs under the influence of the aforementioned risks.

RESEARCH CONTRIBUTIONS

Theoretically, the research provides additional contributions to strengthen the existing theoretical framework on credit risk, liquidity risk, and bank stability by inheriting and synthesizing relevant definitions, measures, and theories Particularly, meticulously synthesized measures from a range of studies serve as the basis for effectively selecting key variables in the research model, thereby ensuring that the research outcomes are contextually relevant and reliable within the context of Vietnam’s JSCBs

Empirically, the research utilizes a highly updated data set spanning from

2013 to 2023 to provide the following significant empirical contributions

Firstly, the research supplements empirical evidence regarding the effects of credit risk and liquidity risk on the stability of Vietnam’s JSCBs, especially considering the interactive effect alongside the individual effects of these two risks This is vital for domestic JSCBs, where these risks are inevitable in core banking activities like deposit mobilization and credit extension Moreover, the potential interaction between these risks may occur when banks engage in deposit mobilization to facilitate credit extension, as stated in Section 1.1

Secondly, the research provides additional empirical evidence on the effects of bank-specific factors and macroeconomic factors on the stability of Vietnam’s JSCBs, thus ensuring the comprehensiveness of the research as bank stability is not just influenced by typical bank risks

Thirdly, based on the empirical findings, the research offers recommendations to Vietnam’s JSCBs to ensure their stability under the influence of credit and liquidity risks, bank-specific factors, and macroeconomic factors This is particularly crucial in the context of developing countries like Vietnam, as the stability of the domestic banking system serves as a vital driver for economic growth, as discussed in Section 1.1.

RESEARCH STRUCTURE

To achieve the research objectives outlined in Section 1.4, the research is structured into five chapters, each with its own specific targets to be attained, as presented below

Chapter 1 presents a comprehensive introduction to the research through the reasons for the research, research objectives, research subjects and scope, research methodology, research contributions and research structure This serves as the basis for the research contents in subsequent chapters.

LITERATURE REVIEW

DEFINITIONS

For most studies in the field of finance, risk is understood as the uncertainty of future outcomes (Reilly & Brown, 2012) This implies that these outcomes can be either favorable or unfavorable, in other words, risk encompasses both positive and negative aspects In line with this perspective, Hull (2012) suggests that risk can lead to unforeseen losses and damages, but it can also bring about unexpected opportunities However, in the context of the banking sector, Tram Thi Xuan Huong (2013) asserts that risk is associated with losses, damages, or other related factors, resulting in unforeseeable challenges and difficulties Associating risk with loss and damage is particularly appropriate when discussing credit risk and liquidity risk, reflecting a prudent approach to bank risk management as these are the most critical risks determining banks’ survival (Imbierowicz & Rauch, 2014) Below is the theoretical framework concerning the aforementioned risks

For the Vietnamese system of JSCBs, which operates with a profit-oriented goal, credit activities remain essential, contributing the highest proportion of income to the bank's total operating income (Nguyen Minh Ha & Nguyen Ba Huong, 2016) However, these activities inherently involve credit risk Accordingly, this risk is characterized by several key definitions outlined below

According to the Basel Committee on Banking Supervision (2000), credit risk can be simply described as the potential that a borrower or counterparty associated with a bank may not fulfill its obligations as per the agreed terms Additionally, credit risk implies a loss of income for the bank, thus credit risk can also be understood as the loss of bank income due to the borrower's delay in making timely or full payments as stipulated in the credit agreement (Ahmed & Khan, 2007) Furthermore, Gestel & Baesens (2008) specify that credit risk arises when a borrower either cannot repay the debt or repays it late due to default or intentional breach of agreed terms

In Vietnam, clause 1, Article 3 of Circular 11/2021/TT-NHNN issued by the SBV on July 30, 2021, defines credit risk within banking activities is the potential for financial loss stemming from a borrower’s incapacity to repay debts, either partially or fully, to a bank under a contract or established between the borrower and the bank Accordingly, Bui Dieu Anh (2020) mentions that the debt to be repaid by the customer includes both principal and interest, thus stating that credit risk is the possibility of loss to a bank when its borrower breaches the credit commitment: fails to fulfill or fulfill incompletely the obligation to repay the principal and interest to the bank

Thus, credit risk essentially includes the following implications: (i) debt repayment ability – the inability of the borrower to pay part or all of the debt reflects credit risk; (ii) debt structure – the inability of the borrower to repay both principal and interest implies credit risk; (iii) repayment term – the failure of the borrower to repay the debt on time according to the contract or agreement with the bank indicates credit risk In summary, credit risk refers to the likelihood of capital loss for the bank when the borrower loses the ability to repay debt in part or in full, including both principal and interest, and/or fails to make payments within the time specified in the credit agreement between the bank and its borrower Moreover, the aforementioned definitions imply that the causes of credit risk originate primarily from the borrower, highlighting this as the main cause of such risk (Le Thong Tien et al., 2022), alongside causes from the bank (inappropriate credit policy; capability of credit risk management not keeping pace with credit growth rate; poor-quality liquid assets) and unavoidable external factors (epidemics, natural disasters, wars, economic conditions and policies) The corresponding consequences include increased operating costs, loss of operating income, resulting in reduced profitability and asset size, diminished bank reputation, and if prolonged at a high level without control, leading to bankruptcy

Given the associated consequences outlined in Section 2.1.1.1, credit risk must be quantified to ensure rigorous control, thereby preventing loose management that could jeopardize long-term bank stability Below are some of the representative measures widely used in both domestic and foreign research literature

The non-performing loan (NPL) ratio is a traditional and widely-used measure of credit risk to date (Fernández et al., 2016; Lachaab, 2023; Nier & Baumann, 2006) According to clauses 8 and 9 of Article 3 of Circular 11/2021/TT-NHNN, NPLs are defined as loans classified into groups 3 (sub-standard loans), 4 (doubtful loans), and

5 (loss loans), excluding off-balance sheet commitments Accordingly, the NPL ratio is the ratio of NPLs to the total loans from groups 1 to 5 (or total gross loans) This ratio is used to proxy for credit risk in several studies (Dang Van Dan & Dang Van Cuong, 2021; Lachaab, 2023; Nguyen Quoc Anh, 2023; Setiawan et al., 2019) Specifically:

Non-performing loan ratio = Non-performing loans

Total gross loans (2.1) Therefore, as NPLs are those in the three lowest debt groups out of five, including loans that are overdue for principal and/or interest payments of more than

90 days, and have a high likelihood of bank capital loss despite being rescheduled or extended of repayment terms, the NPL ratio clearly reflects the credit quality, the effectiveness of credit policy, the difficulty in capital recovery, and the risk of capital loss for the bank Essentially, a higher proportion of non-performing loans in total gross loans implies higher credit risk, lower credit quality, and less effective credit policy In addition, a high NPL ratio may indicate the bank's relatively high-risk appetite in pursuit of high profits, thereby necessitating the development of credit risk management policy that aligns with the level of bank risk tolerance

Moreover, the loan loss provision (LLP) ratio is also a relatively common measure of credit risk, alongside the NPL ratio According to clause 3, Article 3 of Circular 11/2021/TT-NHNN, risk provisions are amounts of money set aside and recorded as operating expenses to cover potential risks associated with bank debts Accordingly, the LLP ratio is the ratio of loan loss provisions to the total gross loans This ratio is a representative indicator of credit risk in various studies (Nguyen Quoc Anh, 2023; Nguyen Quoc Anh & Duong Nguyen Thanh Phuong, 2021; Serwadda, 2018) Specifically:

Loan loss provision ratio = Loan loss provisions

Total gross loans (2.2) Therefore, LLPs fundamentally represent expenses that banks must set aside to reserve for potential losses associated with their loans, based on the classification of debts into five groups according to Circular 11/2021/TT-NHNN, thus comprehensively measuring credit risk by regarding all five debt groups Accordingly, as LLPs constitute a higher proportion of total gross loans, it reflects higher credit risk due to deteriorating bank credit quality, thus prompting banks to increase provisions to enhance prudence and proactivity in credit risk management Concurrently, this also enhances the bank's ability to cope with credit risk in cases of non-recoverable loans, which need to be addressed using available LLPs

In addition to the two fundamental measures of NPL ratio and LLP ratio, credit risk is also quantified by the loan-to-asset ratio and the credit growth rate (measured by the change in total loans at a certain point in time compared to a previous point in time) These measures represent credit risk in numerous studies (Adusei, 2015; Hakimi et al., 2017; Nguyen Duc Trung et al., 2022; Nguyen Minh Ha & Nguyen Ba Huong, 2016) Specifically:

Loan-to-asset ratio = Net loans

Total assets (2.3) Credit growth rate = Loans t - Loans t-1

Therefore, all measures above reflect bank credit size, wherein the loan-to- asset ratio indicates the scale of loan assets within the bank's asset portfolio while the credit growth rate signifies the bank’s credit scale over time Although these measures do not directly reflect credit risk, a rapid expansion of bank’s credit scale without alignment with the bank's risk management capacity may increase credit risk, thus posing a direct threat to long-term bank stability

Fundamentally, liquidity refers to the ease of converting a specific asset into cash when needed and the market is still able to accept such transactions (Duttweiler, 2011) Specifically, liquidity is the ability of an asset to be quickly bought or sold, with reasonable costs and prices without significantly affecting the asset’s price (Nguyen Kim Chi et al., 2023) However, from the standpoint of banks, the Basel Committee on Banking Supervision (2008) states that liquidity represents a bank's ability to finance asset growth and fulfill obligations as they come due without incurring losses that are considered unacceptable

THEORIES

With the research topic on the effects of credit risk and liquidity risk on the stability of Vietnam’s JSCBs, the research conducts an review of related theories and identifies some prominent theories below

Theory of asymmetric information of Akerlof (1970) states that market failures can occur due to disparities in the information held by transaction participants in markets This theory is particularly applicable to various markets, including the credit market In lending activities, as borrowers inherently possess more information about their loans (such as borrowing need, goodwill, and debt repayment ability) than banks, information asymmetry arises, leading to two main consequences: adverse selection and moral hazard Adverse selection occurs when banks make erroneous credit decisions by lending to high-risk borrowers based on incomplete and inaccurate information Moral hazard, on the other hand, arises after loans have been disbursed when borrowers use the borrowed funds for purposes other than those stipulated in the credit agreements, believing that responsibilities for their actions are shared with banks Thus, if banks fail to accurately assess the risk level of borrowers and/or do not strictly monitor how funds are utilized post-disbursement, they face a high likelihood of capital loss and accordingly high credit risk, hence seriously threatening bank stability This is because lending activities are key business activities, serving as the primary source of income that reinforces bank stability In addition, if capital is indeed lost, the bank liquidity decreases as lost capital is essentially a bank debt that must be timely repaid to depositors, leading to increased liquidity risk and further impacting bank stability negatively Therefore, this theory highlights the importance of information in lending activities, demanding that banks upgrade their information systems and boost credit monitoring practices to mitigate credit and liquidity risks arising from information asymmetry in lending activities to ensure bank stability

The financial instability hypothesis introduced by Minsky (1977) posits that financial instability originates during periods of economic growth when the risk of financial crises is not recognized According to this theory, financial instability is cyclical and closely linked to financial crises caused by excessive credit growth within the banking system The focal point of this theory is the “Minsky moment” – the point at which the financial system transitions from stability to crisis through three stages: hedge finance, speculative finance, and Ponzi finance

In the first stage, as the economy has just stabilized in the post-crisis period, economic agents use their own capital or borrow cautiously from banks to ensure full repayment of principal and interest with their available income In the next stage, as the economy booms, economic agents borrow more from banks to invest for profits, expecting to repay interest from returns of investment assets but struggling with continuous principal repayments At this point, banks expand their credit scale, accepting higher credit risk to increase profits, thereby embedding a risk of bank instability The final stage is Ponzi finance, where banks lend without considering borrowers' repayment capacity, hoping values of investment assets will rise enough for borrowers to sell and repay the banks At this point, borrowers' repayment capacity declines sharply due to reliance on the appreciation of investment assets rather than their returns, leading to increased credit risk for banks Concurrently, banks face difficulties in capital recovery, diminishing the ability to meet debt obligations and increasing liquidity risk, which forces them to liquidate assets to maintain their liquidity

Consequently, credit risk and liquidity risk escalate simultaneously, significantly reducing bank’s asset scale and increasing the risk of bankruptcy, thereby amplifying bank instability The end of the Ponzi stage marks the beginning of an economic recession Minsky (1977) thus emphasizes the importance of the legal framework and government intervention to mitigate the adverse impacts when the Minsky moment occurs

The credit rationing theory developed by Stiglitz & Weiss (1981) underscores the inherent risks in banking operations According to this theory, banks may not always use interest rates to balance supply and demand in the credit market due to asymmetric information between banks and borrowers This issue arises when borrowers have better knowledge of their repayment ability than banks, leading to adverse selection (making erroneous credit decisions) for banks and moral hazard among borrowers Consequently, banks may not accurately assess the level of credit risk and the repayment intentions of borrowers, resulting in higher lending rates and increased collateral requirements to compensate for this risk

As a result, high-risk borrowers are willing to accept higher interest rates, while low-risk borrowers face difficulties in accessing bank funds, accordingly increasing the proportion of high-risk loans in the bank's loan portfolio and deteriorating its quality In addition, high borrowing costs may prompt borrowers to change their capital usage, investing in riskier projects to maximize returns and ensure their repayment ability, thereby exposing banks to a higher risk of capital loss as they bear most of the risk as the capital provider

Consequently, increased credit risk leads to higher liquidity risk because banks may be unable to recover funds promptly to meet debt obligations at maturity, hence increasing the risk of bank insolvency and bank instability Therefore, Stiglitz & Weiss (1981) propose that banks should create credit restrictions by limiting the number of loans rather than limiting loan sizes or increasing interest rates to effectively manage credit and liquidity risks, thereby minimizing bankruptcy risk and reinforcing bank stability.

RESEARCH METHODOLOGY

RESEARCH HYPOTHESES

To examine the effects of credit risk and liquidity risk on the stability of Vietnam’s JSCBs, the research formulates hypotheses based on the research questions outlined in Section 1.3 to achieve the stated research objectives

Hypothesis 1: Credit risk exerts a negative effect on the stability of the Vietnam’s JSCBs

Credit risk fundamentally reflects the potential capital losses a bank may face when borrowers either partially or wholly fail to make repayments, and/or when such payments are not made on time as per agreements This negatively impacts bank’s capital, disrupting its primary economic function of allocating resources between entities with capital surpluses and those with deficits within the economy When bank capital is impaired, its ability to repay funds to capital-surplus entities that previously provided capital diminishes, whereas the credit demand from capital-deficient entities remains urgent This undermines the bank's function of resource allocation, implying bank instability according to the definition of bank stability discussed in Chapter 2, thereby suggesting that credit risk has a negative effect on bank stability This perspective is supported by the financial instability hypothesis of Minsky (1977) and the credit rationing theory of Stiglitz & Weiss (1981), which suggest that excessive credit growth or increased lending rates intended to mitigate information asymmetry also heighten credit risk, leading to greater bank instability Several empirical studies affirm this perspective, despite differences in research samples and measures of credit risk and bank stability (Khan & Yilmaz, 2022; Le Thong Tien et al., 2022; Nguyen

Duc Trung et al., 2022; Nguyen Quoc Anh & Duong Nguyen Thanh Phuong, 2021; Sunarsih et al., 2022)

Hypothesis 2: Liquidity risk exerts a negative effect on the stability of the Vietnam’s JSCBs

Fundamentally, bank liquidity reflects its ability to timely meet increasing asset needs and financial obligations when due If a bank loses or experiences a reduction in this capacity, it leads to increased capital-raising costs, indicating liquidity risk for the bank Bank liquidity also pertains to bank solvency; a deterioration or loss in this capability reveals a decline in the bank's ability to perform its core economic function of allocating resources within the economy as a financial intermediary Consequently, if this function is not smoothly executed, the bank becomes more unstable as defined in the definition of bank stability in Chapter 2, leading to the assertion that liquidity risk has a negative effect on bank stability This assertion is reinforced by the financial instability hypothesis of Minsky (1977) and the credit rationing theory of Stiglitz & Weiss (1981), indicating that excessive credit growth and higher lending induce both credit and liquidity risks as banks face capital losses and difficulties in fund repayments for their customers This situation may escalate demands for massive capital withdrawals from customers, worsening liquidity issues for banks and potentially leading to bankruptcy Various empirical evidence supports the viewpoint that liquidity risk adversely affects bank stability, although measured differently (Hakimi et al., 2017; Nguyen Duc Trung et al., 2022; Phan et al., 2019; Sunarsih et al., 2022)

Hypothesis 3: The interaction between credit risk and liquidity risk exerts a negative effect on the stability of the Vietnam’s JSCBs

Given that the individual effects of credit risk and liquidity risk are considered negative for bank stability, the interaction between these two risks may fundamentally make banks even less stable In other words, credit risk and liquidity risk may have an interactive negative effect on bank stability, which not only reflects the individual adverse effects of two risks but also demonstrates a strong link between them in banking operations Specifically, when credit risk arises from borrowers' inability to repay loans and/or to make timely repayments, the bank faces capital losses As a financial intermediary, the bank channels capital between capital-surplus and capital- deficient entities in the economy Hence, the capital losses involve funds mainly mobilized from capital-surplus entities that the bank is obligated to repay Consequently, the emergence of credit risk reduces the bank's ability to meet its financial obligations, which correspondingly increases liquidity risk, further threatening bank stability This demonstrates that the interactive negative effect of both risks is significant for bank stability The perspective of a substantial negative relationship between the interaction of these risks and bank stability is further supported by the financial instability hypothesis of Minsky (1977) and the credit rationing theory of Stiglitz & Weiss (1981), along with empirical evidence from Ghenimi et al (2017), Hakimi et al (2017) and Khan & Yilmaz (2022).

RESEARCH MODEL

The construction of a research model is crucial for validating research hypotheses, thereby answering research questions to achieve the set objectives Below is the model and a detailed description of its variables

To examine the individual and interactive effects of credit risk and liquidity risk on bank stability, the research builds upon the empirical model of Ghenimi et al (2017), concurrently makes adjustations for model variables to represent the operational environment of Vietnam’s JSCBs from 2013 to 2023 The general research model is as follows:

- Dependent variable: Y i,t as stability of bank i at time t

- Independent variables are represented as follows:

Y i,t-1 : The one-period lag of the dependent variable

X i,t : Risks of bank i at time t, including: CR, LR and CRxLR

Z i,t : Bank-specific control variables of bank i at time t, including: CAP, EFF, ROA, SIZE

M t : Macroeconomic control variables at time t, including: GDPG và INF ε i,t : Error term

The specific research model is structured as follows: lnZ-score i,t = β 0 + β 1 lnZ-score i,t-1 + β 2 CR i,t + β 3 LR i,t + β 4 CR i,t xLR i,t + β 5 CAP i,t + β 6 EFF i,t + β 7 ROA i,t + β 8 SIZE i,t + β 9 GDP t + β 10 INF t + ε i,t (3.2)

- Dependent variable: lnZ-score i,t , referring to bank stability

- Independent variables are represented as follows: lnZ-score i,t-1 : Bank stability in the previous year

CRxLR: Interaction between credit risk and liquidity risk

- i and t subscripts: individual bank (26 banks) and year (11 years from 2013 to 2023), respectively

- β 1 ,β 2 … ,β 10 : Coefficients of regression for independent variables

- ε: Error term to capture the variation in dependent variable not explained by independent variables

3.2.2 Description of variables in research model

Regarding bank stability, the research employs the Z-score as a metric to gauge the overall stability of banks This score, essential and easily accessible through bank’s accounting data, is well-regarded in the academic community (Hassan et al., 2019) for its robust predictive power regarding bank stability, with the ability to forecast a 76% probability of bank failures (Chiaramonte et al., 2015)

The Z-score simultaneously assesses bank profit (ROA), financial leverage (equity-to-asset ratio), and profit volatility (standard deviation of ROA) to gauge a bank's insolvency risk It measures how many standard deviations the actual ROA must fall below its expected value before the bank equity is exhausted, leading to complete insolvency (Boyd et al., 1993) Consequently, a higher Z-score signifies enhanced bank stability, as it inversely indicates the likelihood of bank default, calculated as follows:

TA i,t σ(ROA) i,t (3.3) Where: Z-score i,t , ROA i,t , (E TA⁄ ) i,t , and σ(ROA) i,t refers to bank stability, return on assets, equity-to-asset ratio and standard deviation of ROA of bank i in year t, respectively Due to the high skewness of Z-scores among the banks in the research, the natural logarithm of the Z-score is used to smooth these values (Ghenimi et al., 2017; Laeven & Levine, 2009) In addition, the standard deviation of ROA is calculated based on the bank's accounting data over a three-year period from t to t-2 (Gupta & Kashiramka, 2024)

▪ Bank stability in the previous year

Bank stability exhibits a time-persistent characteristic (Bermpei et al., 2018)

In other words, past values of the Z-score (which represent bank stability as discussed above) are likely to determine its current values, implying the autoregressive nature of the Z-score (Gupta & Kashiramka, 2024) Thus, when constructing the regression model (3.2), the unique characteristics of the Z-score are considered by incorporating its one-period lagged value, more specifically the bank stability in the previous year, to mitigate estimation biases due to variable omission, thereby enhancing the robustness and efficiency of the estimators applied for the research model Similarly, the previous year's bank stability, also approached in the form of natural logarithm, has been shown to have a positive impact on current bank stability in several studies (Dao Le Kieu Oanh et al., 2021; Ghenimi et al., 2017; Lachaab, 2023) Therefore, the bank stability in the previous year is expected to positively correlate with current bank stability

Credit risk is represented by the ratio of non-performing loans to total bank gross loans Compared to other risk measures detailed in Section 2.1.1.2, the NPL ratio remains a direct and traditional measure of credit risk (Fernández et al., 2016; Nier & Baumann, 2006) In addition, credit risk proxied by NPL ratio achieves high reliability when calculated based on the classification of bank loans into five groups according to Circular 11/2021/TT-NHNN and NPLs (comprising groups 3, 4, and 5) clearly disclosed in the bank's audited consolidated financial statements over the years Accordingly, a higher NPL ratio indicates a greater proportion of NPLs in the loan portfolio, lower credit quality, and higher credit risk, threatening bank stability (Lachaab, 2023) Furthermore, an increasing NPL ratio also raises the cost of loan loss provision, leading to reduced bank profits and lower bank stability (Nguyen Quoc Anh & Duong Nguyen Thanh Phuong, 2021) Therefore, credit risk is expected to negatively correlate with bank stability

Credit risk (CR) = Non-performing loans

Where: Non-performing loans refer to bank loans that are classified into groups 3, 4, and 5 according to Circular 11/2021/TT-NHNN, and total gross loans refer to total loans before provision for loans

Liquidity risk is measured by the ratio of liquid assets to total bank assets, where high liquidity assets include cash, deposits at the SBV and other credit institutions, as approached by Nguyen Kim Chi et al (2023) to align with the context of Vietnam’s JSCBs This measure, easily calculated from bank’s accounting data, is widely used in numerous studies focusing on liquidity risk originating from bank assets rather than its capital According to Lachaab (2023), the overall stability of a bank may be affected if it lacks assets that can be quickly converted into cash at a reasonable cost to meet immediate financial obligations, such as unexpected deposit withdrawals from customers (Ghenimi et al., 2017) In other words, a higher ratio indicates better bank liquidity, lower liquidity risk, and thus mitigates the risk of liquidity crises, thereby bolstering bank stability Therefore, given that liquidity risk is anticipated to have a negative effect on bank stability, its measure of liquid assets to total assets ratio is expected to positively correlate with bank stability

Liquidity risk (LR) = Liquid assets

Total assets (3.5) Where: Liquid assets include cash, deposits at the SBV and deposits at credit institutions

The interaction between credit risk and liquidity risk (CRxLR)

The interaction between credit risk and liquidity risk is a product variable of credit risk and liquidity risk, indicating the interaction or interactive effect of these two risks Accordingly, the correlation between this interaction and bank stability is negative, due to the individual negative effects of each risk as examined by Lachaab (2023) and Ghenimi et al (2017), which show that the negative impact of credit risk on bank stability increases with rising liquidity risk and vice versa, while conflicting results are found in the research of Nguyen Kim Chi et al (2023) Considering the individual negative influences of credit risk and liquidity risk as predicted above, the interaction between these two risks is also expected to negatively correlate with bank stability

The CAMELS rating system is recognized as an effective tool for assessing the financial health or overall stability of banks (Ben Lahouel et al., 2024; Nguyen Kim Quoc Trung, 2021) Based on this framework, combined with empirical literature on the same topic, the study selects the following factors as control variables related to bank-specific characteristics, corresponding to the six criteria reflected by the CAMELS method, including: capital adequacy (C), asset quality (A), management (M), earnings (E), liquidity (L), and sensitivity to market risk (S) Among these, the previously mentioned credit risk (CR) and liquidity risk (LR) represent the criteria A and L, respectively Thus, the remaining CAMELS criteria will be presented below

Bank capitalization is measured by the ratio of equity to total bank assets, representing the "C" criterion of the CAMELS framework This ratio reflects the bank's capitalization level and its capital adequacy Theoretically, the bank capitalization ratio is a component used in calculating the Z-score in Formula (3.2) mentioned earlier, suggesting that, under constant profit and profit volatility, a higher capitalization level indicates greater bank stability Furthermore, empirical evidence also supports this positive theoretical relationship Kửhler (2015) highlighted the particularly important role of equity capital in a bank's asset portfolio in contributing to bank stability Specifically, Ghenimi et al (2017) found that banks with high capital adequacy levels are better able to withstand crises, thereby minimizing the risk of bank insolvency and reinforcing bank stability Therefore, bank capitalization is expected to have a positive correlation with bank stability

Cost management efficiency is represented by the ratio of total bank operating expenses to total operating income, reflecting the "M" criterion of the CAMELS framework A higher proportion of operating costs in total operating income indicates less effective bank management, as it shows the bank is not managing its cost items efficiently, adversely affecting banking operations and reducing bank stability (Nguyen Duc Trung et al., 2022) In addition, Lachaab (2023) found that banks with greater cost efficiency are more stable, as a lower ratio facilitates banks to mobilize deposits at lower costs, thereby granting more loans to achieve higher profits, hence leading to greater bank stability Therefore, cost management efficiency is expected to have a negative correlation with bank stability

Cost management efficiency (EFF) = Total operating expenses

Bank profitability is approached by the ratio of net income to total bank assets, which represents the "E" criterion of the CAMELS framework According to Formula (3.2) that measures bank stability through the Z-score, ROA is a component of this index and theoretically, a higher ROA suggests a higher Z-score, indicating that better profitability of the bank leads to its greater stability, assuming constant factors in bank capitalization and profit volatility Accordingly, Khan & Yilmaz (2022) found that ROA consistently correlates positively with bank stability, supporting the aforementioned theoretical positive correlation In addition, Ghenimi et al (2017) noted that the more efficiently banks utilize their assets to generate profits, the higher their creditworthy becomes This, in turn, facilitates the expansion of credit scale to increase profits, thus enhancing bank stability Therefore, bank profitability is expected to have a positive correlation with bank stability

Bank profitability (ROA) = Net income

Bank size is measured by the natural logarithm of total bank assets to minimize data dispersion due to high skewness in total assets’ distribution across banks, also representing the "S" criterion of the CAMELS framework Accordingly, larger banks, in terms of total assets, are found to have higher stability compared to smaller banks (Phan et al., 2019), indicating that bigger banks are more stable In addition, Dao Le Kieu Oanh et al (2021) argued that larger banks can leverage economies of scale to positively impact bank stability; in other words, larger banks have lower average operating costs and better capability to diversify asset portfolios to minimize overall bank risk, thus achieving higher profits with lower overall risk, leading to greater bank stability Therefore, bank size is expected to have a positive correlation with bank stability

Bank size (SIZE) = ln(Total assets) (3.9)

To control for the macroeconomic environment in banking operations, the research includes two factors: GDP growth and inflation, in the empirical model (3.2)

RESEARCH DATA

With the research objective of examining the effects of credit risk and liquidity risk on the stability of Vietnam’s JSCBs from 2013 to 2023, the research gathers secondary data on the two main groups of factors below

For bank-specific factors, the research retrieves data on the financial situation of Vietnam's JSCBs over 11 years from the banks' financial reports through the FiinPro-X database The selected banks must meet the criteria below

Firstly, the banks are listed on the Ho Chi Minh Stock Exchange (HOSE), Ha

Noi Stock Exchange (HNX) and Unlisted Public Company Market (UPCOM)

Secondly, the annual bank financial statements must be consolidated, audited, fully and transparently published over the years within the research period

Accordingly, the research select 26 out of 31 Vietnam’s JSCBs that meet two criteria above (details in Appendix 1), representing approximately 84% of the Vietnam’s JSCB system as of March 31, 2024, over an 11-year research period, thereby producing a panel data set with a sample size of 286 observations

For macroeconomic factors, the research retrieves data on Vietnam's macroeconomic environment over the same 11-year period through the World Bank database.

RESEARCH RESULTS AND DISCUSSION

DESCRIPTIVE STATISTICS

Descriptive statistical analysis is conducted for all variables of the research model, including the dependent variable and independent variables, as shown in the five statistical criteria in Table 4.1 below

Table 4.1 Descriptive statistics for variables in the research model

Variables Number of observations Mean Standard deviation Min Max

Source: Author’s computation using Stata 17

Table 4.1 shows 286 observations for all variables, corresponding to the dataset comprising specific characteristics of 26 banks over 11 years from 2013 to

2023, indicating that the research data was fully collected without any missing values

Bank stability is measured by the natural logarithm of Z-score The lnZ-score has an average value of 4.0201, with a maximum of 8.2747 and a minimum of 1.3949 Correspondingly, Z-score averages at 120.8575, with a peak of 3923.3340 and a low of 4.0344 An average Z-score of 120.8575 indicates that most JSCBs operated relatively stably during the 2013-2023 period This is because ROA would need to decrease by approximately 121 times standard deviation from its mean for a bank to become insolvent due to the depletion of its equity, leading to bankruptcy However, the standard deviation of Z-score is 283.1233, significantly higher than its mean, indicating that the Z-score values are widely dispersed around the mean This suggests considerable variation in the stability levels of JSCBs over the years within the study period, with values ranging from 4.0344 (TPB in 2015) to 3923.3340 (SHB in 2018)

Independent variable on bank stability in the previous year

Bank stability in the previous year (lnZ-score(t-1)) reflects the 1-year lagged value of dependent variable lnZ-score Regarding the original value, Z-score(t-1) has an mean of 118.0004, a maximum of 3923.3340 (SHB in 2018), and a minimum of 4.5663 (NVB in 2022)

Independent variables on bank risks

Credit risk (CR), proxied by NPL ratio, has a mean of 2.23%, ranging from 0.47% (TCB in 2020) to 29.76% (NVB in 2023) This wide range indicates significant differences in credit risk management capabilities among JSCBs over years In addition, liquidity risk (LR), represented by the ratio of liquid assets to total assets, averages 13.90%, with a minimum value of 1.38% (BAB in 2016) and a maximum value of 41.62% (KLB in 2021), highlighting notable differences in bank liquidity among JSCBs over the years within the study period Accordingly, NVB's high NPL ratio of 29.76% in 2023 indicates its relatively poor credit quality, leading the bank to undergo comprehensive restructuring to effectively address its bad debts Conversely, KLB prioritized accumulating liquid assets to 41.62% of total assets in

2021 to bolster bank liquidity, stabilizing banking operations amid the severe outbreak of the COVID-19 pandemic

Independent variables on bank characteristics

Bank capitalization (CAP) has a maximum value of 23.84% (SGB in 2013), a minimum value of 4.06% (BID in 2017), and a mean value of 8.78% In addition, cost management efficiency (EFF) averages 51.38%, ranging from 22.71% (SHB in 2022) to 172.25% (NVB in 2023) Bank profitability (ROA) fluctuates from -0.72% (NVB in 2023) to 3.58% (TCB in 2021), with a mean of 0.95% Overall, the standard deviations of CAP, EFF, and ROA are relatively high, at 3.21%, 16.15%, and 0.74%, respectively This indicates that the values of these variables are widely dispersed around their means, reflecting significant differences in capitalization level, cost management efficiency, and profitability among JSCBs over the study period Notably, EFF, with a maximum value of 172.25%, indicates that operating expenses far exceeded operating income, and ROA, with a minimum value of -0.72% due to negative net income, were both recorded at NVB in 2023 This suggests that while the bank's expenses continued to accrue, its main income from credit activities was insignificant due to excessive bad debts, leading to the erosion of net income and correspondingly a negative ROA for NVB in 2023 Furthermore, bank size (SIZE) is measured by the natural logarithm of total assets to mitigate the variability in size among JSCBs Thus, SIZE values are primarily distributed around the mean of 32.7056, corresponding to an average actual total asset value of VND 306 trillion

Independent variables on macroeconomic environment

During the 2013-2023 period, Vietnam’s GDP growth (GDPG) averaged 5.80%, with the lowest growth in 2021 at 2.58% and the highest in 2022 at 8.02% Meanwhile, inflation (INF) averaged 3.20%, with the lowest rate of 0.60% in 2015 and the highest rate of 6.59% in 2013.

CORRELATION ANALYSIS

Correlation analysis is conducted to identify the presence of multicollinearity among the independent variables in the research model, using the correlation matrix and Variance Inflation Factor, presented in Table 4.2 below The correlation matrix shows the linear relationships between the independent variables through correlation coefficients for all pairs of independent variables, while the VIF values indicate the degree of correlation between one independent variable with the rest in the research model

Table 4.2 indicates a low correlation among most independent variables, with the correlation coefficients generally ranging from -0.5 to 0.5 However, there are significant negative correlations between certain pairs of variables, such as ROA and EFF (-0.6961), SIZE and EFF (-0.5677) Notably, interactive variables, fundamentally formed by the product of individual independent variables, can inherently cause structural multicollinearity (Gupta & Kashiramka, 2024), as evidenced by a relatively high correlation coefficient of 0.7906 between CRxLR and

CR However, one of the research objectives is to examine the interactive effect of credit and liquidity risks on bank stability, thus the research cannot eliminate this interactive variable to mitigate potential multicollinearity Attempts to alter the measurement of the CR variable are also ineffective, leading to the acceptance of high correlation between the interactive variable and its component This decision is justified considering: (i) all correlation coefficients in the correlation matrix are below 0.8; (ii) all VIF values are less than 10, specifically below 8, and a mean VIF of 3.15 Therefore, the model can be concluded to exhibit non-severe multicollinearity, and the independent variables used are appropriate, allowing the study to proceed with further research steps

Table 4.2 Correlation between variables in the research model

Variables lnZ-score lnZ-score

(t-1) CR LR CRxLR CAP EFF ROA SIZE GDPG INF lnZ-score 1 lnZ-score

Source: Author’s computation using Stata 17

REGRESSION ANALYSIS

The regression analysis is conducted to identify the relationships between the independent variables and the dependent variable of the research model To ensure the consistency of the estimation results for these relationships, the research follows a sequence that begins with employing basic estimation methods to determine the most suitable estimator for the model, subsequently testing for defects in the model with the chosen estimator, and finally deploying alternative estimation method to rectify any detected defects

4.3.1 Regression by Pooled OLS, FEM and REM methods

To initially assess the effects of credit and liquidity risks on the stability of Vietnam's JSCBs, research model (3.2) is estimated using basic panel regression methods: Pooled OLS, FEM, and REM The estimation results and relevant tests to select the optimal model estimator are presented in Table 4.3 below

Table 4.3 Estimation results and estimator selection tests between Pooled OLS,

Variables Pooled OLS FEM REM lnZ-score(t-1) 0.134*** 0.120** 0.125***

***, **, * represent statistical significance at 1%, 5%, and 10%, respectively

Source: Author’s computation using Stata 17

Accordingly, p-value for the F-statistic in the Pooled OLS, FEM, and REM models are all at 0.0000, indicating that these models are statistically significant and suitable for describing the relationships between the independent and dependent variables Moreover, with the Pooled OLS, FEM, and REM estimators, Table 4.3 shows that the research model can relatively well explain the variation of the dependent variable through the independent variables with R-squared values of 66.2%, 61.9%, and 61.7%, respectively The regression results from these three estimators consistently find statistically significant effects at the 1% and 5% levels for most independent variables on bank stability (lnZ-score), except for CR and SIZE However, to determine the most suitable estimator for the model, the research implements F-test and Hausman test as detailed in Table 4.3

Firstly, the F-test is used to choose between the Pooled OLS and FEM estimators for the research model According to Table 4.3, the F-test shows Prob > F

= 0.004, which is less than 5%, leading to the rejection of the null hypothesis that the Pooled OLS estimator is suitable for the research model at the 5% significance level Therefore, the FEM estimator is more suitable for the research model than the Pooled OLS

Secondly, the Hausman test is employed to select between the FEM and REM estimators for the research model According to Table 4.3, the Hausman test shows Prob > chi2 = 0.886, which is greater than 5%, leading to the acceptance of the null hypothesis that the REM estimator is suitable for the research model at the 5% significance level Therefore, the REM estimator is more suitable for the research model than the FEM

Thus, the results of F-test and Hausman test both indicate that the REM method is the most appropriate for estimating the research model (3.2)

4.3.2 Diagnostic tests for model defects by REM method

With REM selected as the estimation method, the research proceeds to identify the presence of defects in the research model, including heteroscedasticity and autocorrelation, through the following diagnostic tests

Table 4.4 Heteroscedasticity test for the research model by REM method

Breusch-Pagan Lagrange multiplier test Chibar2(01) Prob > chibar2

Source: Author’s computation using Stata 17

Heteroscedasticity can undermine the accuracy and reliability of estimation results from the research model As a result, the research employs the Breusch-Pagan test to detect this defect in the model estimated by the REM method According to Table 4.4, the p-value of the Breusch-Pagan test is 0.0025, which is less than 5% This leads to the rejection of the null hypothesis that the REM-estimated model does not exhibit heteroscedasticity at the 5% significance level, confirming the presence of heteroscedasticity in the REM-estimated research model

Table 4.5 Autocorrelation test for the research model by REM method

Source: Author’s computation using Stata 17

Autocorrelation can diminish the accuracy of estimation results from the research model Thus, the Wooldridge test is conducted to detect this defect in the model estimated by the REM method According to Table 4.5, the Wooldridge test reports Prob > F = 0.0000, which is less than 5%, thus rejecting the null hypothesis that the REM-estimated model does not exhibit autocorrelation at a 5% significance level This indicates that autocorrelation is present in the REM-estimated research model

Together with the Breusch-Pagan test, the Wooldridge test concludes that both heteroscedasticity and autocorrelation simultaneously occur in the REM-estimated model Consequently, REM method is no longer appropriate for estimating the research model, thus necessitating the use of an alternative estimator to address these defects and ensure the robustness of the estimation results

4.3.3 Regression by two-step system GMM method

In addition to defects of heteroscedasticity and autocorrelation when applying the REM estimator to the research model, the problem of endogeneity may inherently occur when the dependent variable both affects and is affected by independent variables, leading to biased and inconsistent estimates

Accordingly, the research suggests a reciprocal relationship between credit risk (CR), measured by the NPL ratio and the dependent variable, bank stability (lnZ- score) This relationship is driven by: (i) a higher NPL ratio potentially raising LLP expenses, thus reducing profits and diminishing bank stability; (ii) greater bank stability indicating better capability of credit risk management, which in turn more effectively controls NPLs, leading to a lower NPL ratio Consequently, Durbin and Wu-Hausman tests are conducted as shown in Table 4.6 below to identify endogeneity in the research model

Table 4.6 Endogeneity test for the research model

Source: Author’s computation using Stata 17

Table 4.6 shows that both Durbin and Wu-Hausman tests have p-values of 0.0000, which are less than 0.05, leading to the rejection of the null hypothesis that

CR is not an endogenous variable in the model at a 5% significance level Therefore,

CR is an endogenous variable, indicating that endogeneity exists in the research model

Considering the defects of heteroscedasticity and autocorrelation, as well as potential endogeneity and the dynamic nature of the research model due to the use of lagged dependent variable (lnZ-score(t-1)) as an independent variable, the research decides to apply a two-step system GMM estimator to address these issues Table 4.7 below details the estimation results and related estimator evaluation tests

Table 4.7 Estimation results and estimator evaluation tests for the research model by two-step system GMM method

Variables Coefficient P-value Standard error lnZ-score(t-1) 0.064*** 0.006 0.023

Wald test chi2(10) = 1046685.386 (Prob > chi2 = 0.000)

Hansen test chi2(14) = 16.00 (Prob > chi2 = 0.313)

***, **, * represent statistical significance at 1%, 5%, and 10%, respectively

Source: Author’s computation using Stata 17

To ensure the reliability of the regression results from the two-step system GMM estimator, the research firstly considers the following tests:

Firstly, the number of instruments and groups in Table 4.7 shows that there are 25 instruments, fewer than the number of observed banks (26 groups), thus ensuring the consistency of the estimation results (Arellano & Bond, 1991) and the validity of the test for the overall validity of the instruments (Roodman, 2009)

Secondly, the Wald test in Table 4.7 indicates that Prob > chi2 = 0.000, less than 0.01, thus rejecting the null hypothesis that all estimated coefficients are zero at the 1% significance level In other words, the estimated coefficients for the independent variables in the model are statistically significant, confirming the overall statistical significance of the research model

Thirdly, the Hansen test in Table 4.7 shows that Prob > chi2 = 0.313, greater than 0.1, thus accepting the null hypothesis that instrumental variables are exogenous at the 1% significance level In other words, the instruments used are valid as they are not correlated with the model's error terms, ensuring unbiased and efficient estimates

Lastly, the AR tests in Table 4.7 indicates that the AR(1) test has a p-value of

0.061, below 0.1, thus rejecting the null hypothesis that there is no first-order serial correlation at the 1% significance level Conversely, the p-value for the AR(2) test is 0.213, above 0.1, accepting the null hypothesis that there is no second-order serial correlation at the 1% level Thus, it can be concluded that there is first-order but no second-order serial correlation in the model's residuals

CONCLUSION AND RECOMMENDATIONS

CONCLUSION

The banking system always plays a crucial role in the economic development of a country (Jabra, 2020), thus, bank stability is essential to ensure sustainable economic growth However, banking operations inherently involve various risks, among which credit and liquidity risks are the most significant, determining the survival of banks (Imbierowicz & Rauch, 2014), in other words, potentially posing strong threats to bank stability

Thus, the research “Credit risk, liquidity risk and bank stability: Empirical evidence from Joint Stock Commercial Banks in Vietnam” is conducted using a dataset of 26 Vietnamese JSCBs and Vietnam's macroeconomic factors during 2013 – 2023 The research focuses on the objectives, questions, and hypotheses regarding the individual and interactive effects of credit and liquidity risks on bank stability Accordingly, the research clarifies the following issues:

Firstly, the research constructs the theoretical framework, focusing on the definitions and measures of credit risk, liquidity risk, and bank stability, along with related theories In addition, empirical studies related to the research topic are reviewed to determine the empirical effects of credit and liquidity risks on bank stability

Secondly, the research develops an empirical model and selects appropriate quantitative research methods based on reviewed empirical studies, while also making adjustments to align with the practices of Vietnam’s JSCBs during the research period Accordingly, the research utilizes Stata 17 to implement a two-step system GMM estimator for the research model, thereby obtaining the following notable empirical results

Firstly, considering the individual effects, both credit risk (NPL ratio) and liquidity risk (ratio of liquid assets to total assets) inversely affect bank stability (lnZ- score), meaning that an increase in credit risk or liquidity risk (reduced bank liquidity capacity) reduces bank stability Thus, the first and second research hypotheses concerning the individual negative effects of these two risks on bank stability are accepted

Secondly, considering the interactive effect, the interaction between the two risks positively correlates with bank stability but does not imply a significant increase/decrease in lnZ-score On the one hand, bank liquidity determines the direction of the effect of credit risk on bank stability with a liquidity ratio of 5.80%

In other words, credit risk negatively impacts bank stability when banks’ liquid assets (cash, deposits at the SBV and credit institutions) make up less than 5.80% of total assets; above this ratio, the negative effect of credit risk is eliminated On the other hand, credit risk does not change the negative (positive) relationship between liquidity risk (bank liquidity capacity) and bank stability This means that an increase/decrease in credit risk along with an increase/decrease in bank liquidity (or a decrease/increase in liquidity risk) enhances/reduces the positive effect of bank liquidity, in other words, reduces/enhances the negative effect of liquidity risk on bank stability Thus, the third research hypothesis regarding the negative correlation between the interaction of the two risks and bank stability is rejected due to insufficient empirical evidence The findings on risk interaction reflect the pivotal role of highly liquid assets in strengthening bank liquidity and stability

▪ Bank stability in the previous year lnZ-score(t-1), the one-year lagged value of lnZ-score, correlates positively with lnZ-score This indicates the consistency of bank stability over time, meaning that Vietnam’s JSCBs that are currently stable are likely to become even more stable in the following year This underscores the importance of maintaining high levels of bank stability now to reinforce future stability for banks

Bank capitalization (CAP), profitability (ROA), and size (SIZE) all positively correlate with bank stability, while the negative correlation between cost management efficiency (EFF) and bank stability is not statistically significant In other words, the better capitalized, more profitable, and larger the Vietnam’s JSCBs are, the more stable they tend to be, and vice versa

GDP growth (GDPG) positively impacts on bank stability, indicating that higher economic growth rates in Vietnam create more favorable conditions for Vietnam’s JSCBs to enhance their stability Meanwhile, inflation (INF) has a positive effect on bank stability, but this positive correlation is not statistically significant.

RECOMMENDATIONS

Based on the empirical research findings regarding the effects of credit risk and liquidity risk on the stability of 26 Vietnam’s JSCBs from 2013 to 2023, the research offers the following recommendations for Vietnam’s JSCBs

Firstly, strictly implement credit risk management practices, thereby mitigating negative effect of credit risk on bank stability

Empirical results indicate that increased credit risk, driven by increased NPL ratio, diminishes the stability of Vietnam’s JSCBs Therefore, management practices for credit risk must be rigorously enforced by banks to mitigate this risk through ensuring: (i) Credit activities adhere strictly to legal regulations, banks’ credit processes and credit policies; (ii) Credit portfolios should align with banks’ business strategies, credit scales, credit operation capabilities, and be diversified across borrowers and economic sectors to spread and minimize credit risk; (iii) Provision for credit risk on categorized debts must comply with legal regulations as stipulated in Circular 11/2021/TT-NHNN by the SBV While provision expenses may negatively affect short-term profits, bank stability in the long term is reinforced as banks possess better financial capacity to manage emerging bad debts; (iv) Information system should be regularly upgraded to ensure transparency, diversity, and timeliness of information to minimize potential credit risk due to information asymmetry between banks and borrowers, thereby enhancing the effectiveness of credit decision-making and monitoring

Secondly, expand the size of liquid assets and equity at reasonable levels, thereby strengthening financial capacity to ensure bank stability

Research findings show that increased liquidity risk or reduced bank capitalization decreases the stability of Vietnam’s JSCBs Therefore, to reinforce bank stability, banks need to strengthen their financial capacity by: (i) Increasing holdings of liquid assets (cash, deposits at the SBV and credit institutions) in their asset portfolios with a ratio above 5.80% (as demonstrated by the research) to eliminate the negative effect of credit risk on bank stability In addition, given the inherent and potentially increasing credit risk if credit risk management capabilities do not keep pace with credit growth targets, banks should persist in bolstering bank liquidity by raising the ratio above to mitigate liquidity risk, ensuring bank stability, as joint increases in credit risk and bank liquidity are found to magnify the positive effect of bank liquidity on bank stability Even if credit risk increases due to a higher NPL ratio, timely holding sufficient liquid assets to reinforce financial capacity by liquidating these assets can stabilize credit operations (financing loans and repaying due capital raised for loans), thus maintaining bank stability (ii) Enhancing equity size through boosting stock issuance, profit retention, and attracting capital contributions from strategic investors Accordingly, an increased equity-to-asset ratio will enhance bank financial capacity and capital safety level, reducing bank insolvency risk in the face of financial shocks, thereby strengthening bank stability However, banks need to be mindful to increase ratios above to a reasonable extent so as not to adversely affect bank stability by reducing bank profitability due to the less profitable nature of liquid assets and the high cost of equity capital

Thirdly, reasonably expand the bank size, thereby enhancing bank profitability to strengthen bank stability

Research results indicate that the larger the bank size, the more stable Vietnam’s JSCBs become Therefore, banks should actively boost the expansion of asset size, in other words, increase asset funding through increasing bank capital (raising customer deposits, issuing stocks and bonds, and attracting strategic investors) This capital increase should follow a reasonable timeline, thus allowing banks to achieve dual goals: complying with Basel standards and Vietnamese legal regulations on capital adequacy while enhancing bank stability as the bank size expands Moreover, increasing asset size typically involves a broader bank’s operational network, enabling more effective customer outreach and providing a larger volume of banking products and services, leading to reduced average operating expenses, increased profits, improved profitability and thereby greater bank stability However, banks must also ensure that asset size expansion is accompanied by strict management of the quality and proportion of assets driving this expansion, avoiding an excessive focus on profit objectives with a significant increase in loan assets without proper control over the quality of loan portfolios, causing potential credit risk, and/or maintaining a low proportion of liquid assets due to their low profitability without considering their critical role in ensuring bank liquidity capacity and stability, as highlighted in the second recommendation

Fourthly, sustain the current high level of bank stability, thereby serving as a condition for achieving greater bank stability in the future

Research findings indicate that the stability of Vietnam’s JSCBs is consistent over time Therefore, banks need to ensure the highest possible level of stability now to reinforce future bank stability Accordingly, banks must effectively implement the three previously mentioned recommendations: (i) control and minimize credit risk through strict credit risk management practices; (ii) expand the size of liquid assets (to reduce liquidity risk) and equity relative to total assets to strengthen banks’ financial capability; (iii) increase the bank size while focusing on the quality and proportion of assets that contribute to this growth to enhance bank profitability By ensuring high bank stability now, banks lay the foundation for greater future stability in the future

Finally, develop operational and crisis management strategies adapted to economic development phases, thereby strengthening bank stability

Considering macroeconomic factors, empirical research results show that stronger economic development or higher GDP growth rates lead to greater stability of Vietnam’s JSCBs Since economic growth is beyond the control of individual banks, it necessitates governmental intervention and control, characterized by macroeconomic policy guidance, trade and investment promotion, and human resource development to support sustainable economic development, thereby creating a healthy macroeconomic environment for banks to bolster their stability Meanwhile, banks should proactively enhance their financial strength by increasing liquid assets and equity as stated above, developing business strategies aligned with different economic phases, and formulating strategies to counteract potential adverse economic events, thus ensuring bank stability.

RESEARCH LIMITATIONS AND FURTHER RESEARCH

In addition to providing theoretical and empirical contributions, the research on the individual and interactive effects of credit risk and liquidity risk on the stability of Vietnam’s JSCBs from 2013 to 2023 has certain limitations that need to be addressed through future research orientations as outlined below

Firstly, regarding the research’s temporal scope, the time frame of 2013 –

2023 does not capture the 2007 – 2008 global financial crisis, during which the interactive effect of credit and liquidity risks on bank stability became notably pronounced and partly contributed to bank failures (Imbierowicz & Rauch, 2014)

Therefore, a suggested future research direction is to extend the study period from 2007 – 2023 and include a dummy variable representing this crisis to explore its impact on bank stability This would enhance the comprehensiveness of studies

Secondly, regarding risk measurement in the research, Sections 2.1.1.2 and

2.1.2.2 present various measures for credit and liquidity risks; however, the research now only uses the NPL ratio as credit risk and the ratio of liquid assets to total assets as liquidity risk

Therefore, a proposed future research direction is to employ alternative measures for these risks, specifically the LLP ratio for credit risk and loan-to-deposit ratio for liquidity risk, accordingly adjusting the interactive variable of two risks for research approach diversification This would provide additional empirical evidence on the risk-stability relationship in banks

Thirdly, regarding the measure of bank stability in the research, the Z- score currently only reflects bank profitability (ROA), capitalization (CAP), and profit volatility (standard deviation of ROA) However, Section 2.1.3.2 indicates that there are other variants of the Z-score that could be considered for a more flexible evaluation of bank stability

Therefore, a suggested future research direction is to modify the existing Z- score into its different variants, notably: (i) replacing ROA and its standard deviation with ROE and its standard deviation, where ROE is the ratio of net income to equity (Djebali & Zaghdoudi, 2020); (ii) decomposing the Z-score into its two components, namely risk- adjusted bank profitability (ROA/standard deviation of ROA) and risk- adjusted bank capitalization (CAP/standard deviation of ROA) (Kửhler, 2015) These diverse Z-score approaches would enrich the empirical evidence to evaluate bank stability more comprehensively and effectively

Chapter 5 provides a concise summary of the research objectives, questions, and hypotheses regarding the individual and interactive effects of credit risk and liquidity risk on the stability of 26 Vietnam’s JSCBs from 2013 to 2023, as discussed in Chapters 1, 2, and 3 In addition, empirical research results from Chapter 4 are also concluded, serving as the basis for validating the three research hypotheses and accordingly addressing the first three research objectives concerning the effects of the two risks on bank stability To achieve the final research objective and question, the research draws on the summarized empirical results to offer recommendations for Vietnam’s JSCBs to ensure their long-term stability These recommendations focus on: (i) strict implementation of credit risk management practices; (ii) size expansion of liquid assets and equity to reasonable levels; (iii) reasonable bank size expansion; (iv) maintenance of high current bank stability; (v) development of operational and crisis management strategies aligned with economic development stages Finally, despite positively making theoretical and empirical contributions, the research identifies certain limitations in terms of the research time frame, measurement of bank risks and bank stability in the research; thereby proposing more suitable future research directions

In Vietnam's bank-based financial system, the stability of domestic banks, specifically JSCBs, is a prerequisite for the development of the financial system, thereby fostering sustainable economic growth However, banking activities inherently entail various risks, among which credit risk and liquidity risk pose direct threats to the stability of banks, and the degree of threat is further heightened when these risks interact within banks' credit activities Consequently, the research “Credit risk, liquidity risk and bank stability: Empirical evidence from Joint Stock Commercial Banks in Vietnam” is initiated to identify the individual and interactive effects of credit and liquidity risks on the stability of Vietnam’s JSCBs

Based on this objective, the research formulates research questions and develops corresponding hypotheses based on the theoretical framework regarding definitions and relevant theories along with empirical studies related to the research topic These hypotheses are tested through an empirical model employing suitable quantitative methods on panel data set consisting of 26 Vietnam’s JSCBs from 2013 to 2023 Among various estimation methods – Pooled OLS, FEM, REM and two-step system GMM, through the necessary tests, the research determines the two-step system GMM as the optimal estimator for robust and reliable research results

The empirical findings indicate that credit and liquidity risks individually exert a negative effect on bank stability, thereby validating the first two research hypotheses Meanwhile, the interactive effect of these risks is identified as positive for bank stability, leading to the rejection of the third hypothesis Specifically, an increase in bank liquidity (decrease in liquidity risk) along with an increase in credit risk enhances the positive effect (reduces the negative effect) of bank liquidity (liquidity risk) on bank stability and vice versa Furthermore, to eliminate the individual negative effect of credit risk, the liquidity ratio of banks (the ratio of liquid assets to total assets) needs to exceed 5.80%; below this ratio, credit risk still adversely affects bank stability In addition, considering bank-specific factors, the research finds that the higher the capitalization level, profitability, and expansion of asset size, the more stable Vietnam’s JSCBs become, and vice versa Concurrently, a favorable macroeconomic environment, characterized by higher economic growth rates, also contributes to more stable bank operations Importantly, the research discovers that banks that are stable in the present are likely to become more stable in the future, as current stability determines future bank stability Therefore, with these empirical results, the research provides recommendations for Vietnam’s JSCBs to ensure their long-term bank stability

Bui Dan Thanh & Nguyen Quang Huy (2022) Tác động của sở hữu nước ngoài đến rủi ro thanh khoản tại các ngân hàng thương mại Việt Nam Tạp chí Kinh tế và

Bui Dieu Anh (2020) Giáo trình Tín dụng ngân hàng Nhà xuất bản Kinh tế TP Hồ

Le Dinh Hac, Tran Thi Kim Oanh & Huynh Ngoc Chuong (2017) Phối hợp chính sách tài khóa và chính sách an toàn vĩ mô trong việc ổn định tài chính—Nghiên cứu trường hợp Việt Nam Tạp chí Kinh tế và Ngân hàng châu Á, 141 https://ajeb.hub.edu.vn/vi/article/phoi-hop-chinh-sach-tai-khoa-va-chinh- sach-an-toan-vi-mo-trong-viec-on-dinh-tai-chinh-nghien-cuu-truong-hop- viet-nam

Le Minh Son & Le Phan Quynh Trang (2022, October 15) Ngân hàng Nhà nước đưa SCB vào diện kiểm soát đặc biệt Báo điện tử VnExpress https://vnexpress.net/ngan-hang-nha-nuoc-dua-scb-vao-dien-kiem-soat-dac- biet-4523881.html

Le Ngoc Quynh Anh, Nguyen Quy Quoc & Le Thi Phuong Thanh (2020) Các nhân tố ảnh hưởng đến sự ổn định tài chính của các ngân hàng thương mại Việt Nam Hue University Journal of Science: Economics and Development, 129(5B), Article 5B https://doi.org/10.26459/hueunijed.v129i5B.5845