1152 bank performance and non interest income evidence from vietnamese commercial banks 2023

INTRODUCTION

The urgency of research

In the past few decades, the concentration on banking services has evolved from traditional activities to non-traditional ones (Vivas & Pasiouras, 2010) According to Raluca (2012), globalization and financial deregulation in the banking industry have prolonged a variety of banking goods and services It is undeniable that the traditional role of commercial banks has centered on intermediation and the generation of net interest income through two main activities: collecting deposits on which banks charge interest and issuing loans for which they receive interest income (Craigwell & Maxwell,

2005) However, over the years, commercial banks across the globe have gradually diversified from their conventional roles and sources of income to encompass new pursuits that generate non-interest income (Sherene & Tapper, 2010).

Non-interest income is defined as revenue created by banks from sources other than interest payments, with service fees like ATM fees and loan origination fees serving as the most prevalent examples (Haubrich & Young, 2019) According to the annual reports of banks, non-interest income sources are usually income from service activities, income from investment business and income from other non-interest activities Non-interest income has an advantage over interest income in that it is typically unaffected by legal or regulatory restrictions and is not influenced by economic or financial market cycles (Morris & Regehr, 2014).

Understanding the association between non-interest income and bank performance has imperative implications for business strategy and regulatory policy A favorable correlation between non-interest revenue and bank performance indicates that the shift towards non-traditional activities is beneficial for commercial banks As a result, policies that enable banks to enter new markets and participate in novel, non- traditional off-balance-sheet operations may yield desirable outcomes Moreover, providing a variety of services may also help banks compete with non-financial institutions more effectively However, the relationship between non-interest income and bank performance has not been established with consistency in previous studies.

There are existing literature revealing two contrasting effects of non-interest income on bank performance The proponents argue that there is a reciprocal relationship between non-traditional income sources and banking performance (Smith et al., 2003, Baele et al., 2007 and Lepitit et al., 2008) Theoretically, raising the non- interest income ratio through income diversification could result in more consistent operational income, hence enhancing the bank performance (Chiorazzo et al., 2008; Nguyen et al., 2018) Furthermore, by exploiting managerial skills (Iskandar & McLaughlin, 2007) and economies of scope (Lown et al., 2000), banks can benefit from non-interest activities (Baele et al., 2007; Sanya & Wolfe, 2011; Nguyen et al., 2012; Pennathur et al., 2012; Meslier et al., 2014) Research studies that were carried out in Vietnam have also reached comparable outcomes such as the studies of Le Long Hau and Pham Xuan Quynh (2016), Nguyen Minh Sang and Tran Thi Thanh Tam (2019). The opponents, meanwhile, argued that non-interest income may dilute the comparative advantage of management (Klein & Saidenberg, 2010) and increase earnings volatility (DeYoung & Roland, 2001; Stiroh, 2004; Stiroh & Rumble, 2006) Similarly, Lee et al.

(2014), Senyo et al (2015) and Sun et al (2017) agree that non- interest revenue could increase commercial banks’ exposure to risk Besides, these studies imply that non- traditional activities are likely to make it challenging for banks to raise their income. Therefore, it is apparent that the impact of non-interest revenue on the performance of banks may vary depending on national circumstances and the evolution of the financial system This is one of the research gaps that needs to be filled.

Despite the contradictory empirical findings and theoretical disagreement that have been documented in the literature on economics and finance, it is apparent that the role of non-interest income on bank performance continues to be a topic of intense research However, little is known regarding the effect of this income source on the performance of the banking industry, especially within the context of developing countries, as by far most of those conclusions are either based on developed countries or within the context of non-financial industries According to Hsieh, Lee and Yang

(2013), the majority of current literature evaluations are based on the United States (U.S) or European banks, whereas the relevant topic for Asian banks has not yet induced significant discussion Furthermore, because the banking systems in Asia may differ significantly from those in the USA and Europe in terms of their characteristics and the function that they play in their respective economies, findings from these regions may not necessarily hold for Asian nations This induces that the outcomes of developed nations may not be applicable to developing countries like Vietnam. Therefore, this study was done to provide more in-depth evidence regarding the impact of non-interest income on the performance of Vietnamese commercial banks.

Another intriguing factor that requires further research about the relationship between non-interest income and the performance of the Vietnamese banking system is the current urgent situation in the Vietnamese economy In the development strategy of Vietnam's banking industry to 2025, with orientation to 2030 issued by the Prime Minister and the Vietnamese Government in 2018, the industry development orientation of Vietnam is to increase the proportion of income from non-banking activities in the total revenue to about 16 - 17% Besides, from 2020 to 2021, due to the Covid-19 outbreak, the banking sector has undergone some major changes to meet the requirements of customers and cope with the fierce competition among commercial banks Therefore, analyzing a single market like Vietnam, based on the advantage of a uniform environment, may gain a reference pattern for Vietnam and other emerging markets as well.

In fact, in the light of modernization, the demand for novel services and products is an inevitable trend, especially in the banking and financial sectors If Vietnamese commercial banks can utilize the income produced by non-traditional sources, they could surge their performance and lessen their reliance on core banking activities In order to holistically clarify the impact of non-interest income on the performance of commercial banks in Vietnam, as well as propose some strategies to promote further development of the bank performance, the topic: Bank Performance and Non-interest

Income: Evidence from Vietnamese commercial banks needs to be conducted.

Research objectives

This study intends to investigate the relationship between non-interest income and the performance of Vietnamese commercial banks Particularly, using the bank data source of 26 Vietnamese commercial banks during the period of 2011-2021, the direction and magnitude of the impact of non-interest income on the bank performance will be clarified Subsequently, after a thorough regression measurement, some feasible measures are proposed to enhance the performance of Vietnamese commercial banks.

Following are a few specific objectives that have been set out to achieve the research goals:

(1) Examine the impact of non-interest income on the performance of Vietnamese commercial banks.

(2) Clarify the direction and magnitude of the effect of non-interest income on the performance of Vietnamese commercial banks.

(3) Make some recommendations for Vietnamese commercial banks to improve their performance through noninterest-based banking services.

Research questions

This thesis is aimed at solving below research questions:

(1) How does non-interest income affect the performance of Vietnamese commercial banks?

(2) What is the direction and magnitude of the impact of non-interest income on the performance of Vietnamese commercial banks?

(3) What suggestions should be made for Vietnamese commercial banks to improve their performance through noninterest-based banking services?

Subject and scope of the research

The thesis specifically studies the impact of non-interest income on the performance of Vietnamese commercial banks.

Research time range: The thesis is carried out over the period of 2011-2021 The year 2011 is selected as inflation rate of Vietnam in this year was 18.13 percent, which was the highest proportion since 2008 According to the data from the General Statistics Office of Vietnam, Vietnam also ranked first in the inflation rate among ASEAN nations in 2011 Furthermore, the year 2011 marks the approval and implementation of Decision No 254/QD-TTg of the Vietnamese Prime Minister on

"Restructuring the system of credit institutions during 2011-2015” In addition, the timeline from 2011 to 2021 also includes the year that Vietnam suffered the Covid-19 pandemic, which was in 2020 and 2021 Hence, the thesis focuses on researching the period between 2011 and 2021 to evaluate the economic recovery process of the Vietnam banking sector.

Research spatial range: Due to limitations of data publication of commercial banks, after filtering the sample, this thesis is left with unbalanced panel data for 26Vietnamese commercial banks (out of 31 Vietnamese commercial banks) The sample thus covers about 84 percent of the total Vietnamese commercial banks.

Research methodology

To obtain the target sample, this study manually collects data from the annual financial reports published on the websites of the selected banks as well as on the Vietstock and CafeF websites In addition, this study also extracts macroeconomic data from the World Bank and the General Statistics Office of Vietnam.

The thesis will compare results from empirical analysis with results from previous studies to explain research objectives and research questions.

In this study, quantitative research was used with the support of Stata 17 software To analyze the impact of non-interest income on the performance of commercial banks, Pooled Ordinary Least Squares (Pooled OLS), Fixed Effect model(FEM), Random Effects model (REM) are used Accordingly, the thesis will use F-test and the Hausman test to figure out the most suitable models between FEM and PooledOLS as well as between FEM and REM, respectively Next, with the selected models,tests for heteroskedasticity, autocorrelation and multicollinearity will be conducted If the chosen models incur any issues, the thesis will apply the Feasible Generalized LeastSquares (FGLS) to increase their reliability The study will then use the GeneralizedMethod of Moments (GMM) strategy to resolve the endogeneity issue Finally,Independent Sample T-Test is employed to check the robustness of the research models.

Contribution of the research

Despite the increasing presence of non-interest income at commercial banks,which has become a subject of interest in the industry press and regulatory publications(Feldman & Schmidt, 1999), only a few academic research papers have investigated the effect of increased non-interest income on bank performance (DeYoung & Rice, 2004).While a great deal of research theoretically states that non-interest income will have a favorable influence on the bank's revenue, some pieces offer evidence to suggest the opposite side The various methodologies, analytical approaches, and data sources employed in these investigations will provide different conclusions.

Considering the lack of research on non-interest activities for Asian banks, this thesis intends to deeply analyze the prominence of non-interest income and its effect on the performance of commercial banks in Vietnam Moreover, this research also examines the determinants of other related factors to look at their impacts on overall banking performance Over a number of specifications, the study will set out the most appropriate models representing the impacts of non-interest income on the performance of Vietnamese commercial banks.

A further novel aspect of this study is that it is being undertaken as the world and Asia are recovering from the Covid-19 outbreak Hence, as far as the data is concerned, it will show the effects of the pandemic in the years of 2020 and 2021 In addition, another new and different point of this thesis is that this thesis is conducted after Vietnam implements Decision No 254/QD-TTg of the Vietnamese Prime Minister on "Restructuring the system of credit institutions during 2011-2015," which would indicate how non-interest income has exerted its effect on the Vietnamese commercial banks after the economic recovery process of the banking industry in Vietnam.

This study provides thorough measurements using data during an 11-year period(between 2011 and 2021) of 26 Vietnamese commercial banks to enrich the stream of research on the topic from the perspective of Vietnamese commercial banks Therefore,the research results in this thesis can be used for reference purposes by bank administrators, policymakers, and other stakeholders to contribute to the efficiency of bank administration as well as banking research and governance.

Research structure

The first chapter depicts the background of the research, research question, subject and scope of the research, general methodology, the novelty of the research and research structure.

THEORETICAL BACKGROUND ANDLITERATURE REVIEW .10 2.1 Theoretical of the performance of commercial banks

Bank performance

The banking sector, as a proliferating financial institution, is an active contributor to the economic development of a nation (Babu, 2018; Iskandar et al., 2019) Therefore, banks have become one of the most crucial financial sources of capital for companies (Nimalathasan, 2008) Because of this, the banking system is critical to contemporary society, and its performance is a key indicator of the financial health of a nation. Especially in the present sensitive and competitive market, the performance of commercial banks is likely to be hurt, which would spread consequences for many parties in the economy Additionally, the domino effect brought by the negative externalities of bank failure may represent a systemic risk to the entire banking system, requiring the reshaping and reorientation of bank management, control, and supervision As a result, maintaining the stability of the financial system while also seizing possibilities for business and development are fundamental challenges for the government and commercial banks in this changing environment (Arora & Kaur, 2006).

Commercial banks function as financial intermediation, pooling the funds from lenders and distributing them to borrowers in order to generate profits from the interest spread This supposedly enhances the efficient allocation of capital by channeling funds to those who have a shortage of funds from those who have a surplus (Rose, 2014) Bank performance is determined as the main driver of bank profitability (Nguyen Kim QuocTrung, 2021) It also acts as the pillar and goal of any banking operation (Ferrouhi, 2018).

Banking is a profit-seeking industry In the standard efficiency literature, given the1 output mix and input prices, it is assumed that the bank will select a production plan that maximizes profits or minimizes costs In the situation of perfect competition, economic theories indicate that profit maximization is equal to minimizing costs In practice, however, it can interfere with factors like the modifications in the regulatory framework which would disturb the attainment of desired performance In newer research, bank managers are modeled as maximizing their utility, which is a function of market value and risk These managers may trade off expected return and risk so that production decisions that maximize managers' utility depend not only on the expected profits they make but also on the variability of the profit stream they generate (Hughes, 1999; Hughes, 2000).

The operations of commercial banks are separated into two categories: on-balance- sheet and off-balance-sheet activities (Bui Dieu Anh, 2009) On-balance-sheet activities of banks generally consist of three main services: activities that create capital (bank deposits, borrowing from other financial institutions, investment trusts, etc.); operations related to the application of capital raised by the business of banks (treasury operations, lending, investment activities, etc.); and intermediary service in which commercial banks make payments or other entrustments on behalf of customers to collect fees In particular, the principal activity of a bank is managing the spread between the deposits that it pays consumers and the rate it receives from loans In other words, the bank makes net interest income from the amount of money that comes from the interest rate spread when the interest it receives on assets exceeds the interest it pays on deposits The amount of net interest income a bank generates will depend on a variety of factors, including the quality of the bank's loan portfolio, the average interest rates that each type of loan bears, and whether the loans have fixed or variable rates Also, the ratio of net interest income is known to account for the largest proportion of the total revenue of the bank.

Modifications in operating environments have led banks to become less reliant on traditional models of banking and increase diversity In fact, off-balance-sheet activities have substantially expanded in many banks across the globe, reflecting that banks have been diversifying, making non-interest income a key source of bank revenue in response to increased competitive pressure from non-banking enterprises and international financial2 institutions (Johnson & Murphy, 1987; Reichert, 1985; Heffernan, 2005) In economic terms, off-balance-sheet items are contingent assets and liabilities that affect a commercial bank's financial statements in the future rather than in the present Non- interest income generated by off-balance-sheet activities comprises trading gains and fees, investment banking and net servicing fees, brokerage fees, net gains on asset sales, insurance commissions, fiduciary income, service charges on deposit accounts, net securitization, other foreign transactions, and other noninterest income.

A popular reason for the dramatic growth of bank off-balance-sheet items has been that these activities might have been utilized by banks as a way to boost profits in order to make up for reduced spreads on their typical on-balance sheet corporate lending activity (Nachane & Saibal, 2002) Furthermore, by engaging in off-balance-sheet activities, besides providing high earnings, banks can avoid regulatory fees or taxes because they are not subject to reserve requirements or deposit insurance payments However, the solvency and liquidity of the bank could be impacted by the risks associated with these activities, including those related to market, operational, and credit risks Still, higher interest rates and foreign exchange risk that banks confront in domestic and international markets may be attributed to the large growth in derivatives operations Hence, derivative instruments can be used to control these risks without having to make large changes to the balance sheet.

It is undeniable that the role of banks is of paramount importance Banks operate the payment system, act as a conduit for monetary policy, and are a major source of credit for households, corporations, and governments According to the literature on financial intermediation, commercial banks could solve potential moral hazards and adverse issues resulting from the asymmetric of information between borrowers and lenders by screening and monitoring borrowers As a matter of fact, banks are unique in issuing demandable debt, which confers an informational advantage to banks over other lenders in making loans to informationally opaque borrowers For example, the data obtained from transactions on checking accounts and other sources enables banks to evaluate and manage risk, whereas drafting contracts, and tracking contractual3 performance could address banks’ issues when required.

In the recent past, competition among banks, non-banking financial institutions, and financial markets has intensified Such competition along with technological advancement have resulted in the substantial transformation of banks Therefore, the primary objective of commercial banks is to measure and manage their performance effectively in order to maintain and develop their operations.

2.1.2 Approaches to the bank performance measurement

The metrics used to measure different facets of bank performance have been discovered to be multidimensional (Nickerson & Freeman, 1997; Johnson & Greening, 1999; Rowe & Morrow, 1999; Baum & Wally, 2003; Combs, Crook & Shook, 2005). Despite the rising complexity of banking institutions, the fundamental drivers of their performance remain earnings, efficiency, risk-taking and leverage Hence, there is a multitude of measurements that could be used to assess bank performance, with each group of stakeholders having its own focus of interest Among the vast set of bank performance metrics employed by academics and practitioners, a distinction can be witnessed between traditional, economic, and market-based measures of performance.

The profit dimension is the most widely used measure of bank performance as it evaluates the fulfillment of the firm’s economic goal (De Andres & Vallelado, 2008; Liang et al., 2013) According to traditional measures of performance, return on assets (ROA), return on equity (ROE) are the most widely used Additionally, given the importance of the intermediation function of banks, net interest margin (NIM) is typically monitored These indicators are also in line with the prior studies of Nguyen (2012), Venkatesh & Suresh (2014).

Return on assets (ROA) is computed by dividing the bank’s net income by its total assets, usually the average value over the year ROA reflects the management’s ability to utilize the bank’s financial and real investment resources to generate profits (Hassan &Bashir, 2003) ROA is a common starting point for analyzing earnings because it gives an indication of the return on the bank’s overall activities To put it differently, this ratio4 illustrates the capacity of banking performance while also measuring the efficiency of the company in generating income from asset management The better the bank performs, the higher the ROA because of the higher returns, which leads to an increase in the assets of the company (Dendawijaya, 2001).

„ , , Netincome Return on assets Average total assets

Return on equity (ROE) is calculated by dividing net income by average shareholders’ equity ROE can be used as an indicator to assess management effectiveness in using equity financing to fund operations and grow the company ROE is chosen since: (i) it suggests a straightforward evaluation of the financial return on investment of the shareholders; (ii) it is easily accessible for analysts as it depends simply on public information; and (iii) it enables comparisons between different companies or various economic sectors Nonetheless, in a typical bank, shareholders’ equity is usually small compared to other funding sources used to fund the assets of banks Hence, ROE usually exceeds ROA The bank is highly leveraged and may have limited access to more borrowing if ROE is very high relative to ROA Therefore, the bank performance should be evaluated by return on equity in parallel with return on assets to reach a much more precise conclusion about the profitability, financial efficiency, and capital adequacy of commercial banks.

Return on equity Average total equity

Long-term sustainability is a requirement for the operation of any financial institution The ROA and ROE ratios mentioned above are thus the metrics that are most usually used to evaluate the performance of commercial banks in previous studies (Michel

The net interest margin (NIM) is a proxy for the income generation capacity of the intermediation function of banks This ratio measures the difference between the interest income generated by banks and the amount of interest paid out to their lenders relative to5 the value of their interest-earning assets High NIM indicates a high difference between deposit rates and lending rates (including other interest-earning assets), and vice versa. NIM is calculated by dividing the net interest income by the interest-earning assets It represents the revenue generated by interest-related activities for banks (Berger, 1995; Barajas et al., 1999; Naceur & Goaied, 2001).

Netinterestincome Net interest margin Interest-bearing assets

In recent years, banks have increasingly implemented novel performance metrics based on the concept of economic profit, other than accounting earnings It is undeniable that capital management is the first to be driven by risk As a result, banks have to carefully evaluate the potential unexpected losses linked to each specific activity since risk might trigger losses that deplete their capital Two important metrics that constitute the foundation of risk-capital models are Economic Value Added (EVA) and Risk- Adjusted Return On Capital (RAROC).

Theoretical review of non-interest income

A bank’s shareholders are entitled to its profits, thus maximizing these profits is their topic of interest (Abedifar, 2018) Interest income and non-interest income are the two main revenue sources of commercial banks In addition to traditional bank lending,which usually accounts for the largest proportion of the bank’s total revenue, banks also engage in non-traditional off-balance-sheet services that generate non-interest income.

As discussed by Kohler (2004), non-interest income is the aggregate of various8 sources of income that not only includes fee income and commissions, which are closely related to market-oriented activities such as guarantees and security but also consists of traditional banking-related income such as payment service fees and commission income arising from the sale of insurance and other products In addition, noninterest income is defined as revenue created by banks from sources other than interest payments, with service fees like ATM fees and loan origination fees serving as the most prevalent examples (Haubrich & Young, 2019) According to Hoang Ngoc Tien and Vo Thi Hien

(2010), non-interest income is income from service activities; foreign exchange trading, gold, silver; securities trading and other service activities.

In short, non-interest income includes net gains on trading and derivatives, net fees and commissions, net gains on other securities and other operating income.

2.2.2 Theoretical review on the impacts of non-interest income on the performance of commercial banks

Many studies have noted the impact of non-interest activities on promoting banking performance as well as decreasing bank risks However, there are still opposing views that income diversification by engaging in non-interest activities would make the bank's income streams more vulnerable, thereby reducing the bank performance In fact, banks that decide to increase their noninterest revenue activities are confronted with severe interbank competition as a result of the significant development of financial liberalization and globalization so as to expand, achieve efficiency, and reduce idiosyncratic risk.

Obviously, labor expenditures, administrative costs, and marketing costs make up the majority of the operating expenses for non-interest revenue (Limei, 2016) Therefore,commercial banks have to invest much more money to market new services than those paid for traditional interest-earning activities, especially when new non-interest financial products are released Moreover, asset securitization products of commercial banks could cause asset circulation, increase liquidity, and raise price volatility risk In this circumstance, high-quality administration is required in commercial banks For example, a commercial bank may face considerable investment risk if it lacks investment experience9 when establishing noninterest businesses.

Another problem deriving from this is the volatility that is brought about by noninterest income of commercial banks (Deyoung & Rice, 2001) Within the context of a continuing lending relationship, the primary input required to produce loans is variable (interest expense); in contrast, the main inputs needed to produce more fee-based products are fixed or quasi-fixed (labor expense) As a result, Bourke (1989) states that fee-based activities may necessitate larger operating leverage lending operations, making banks more exposed to a fall in bank revenues In previous literature, some used Portfolio Theory to describe how non-interest revenue affects the performance of banks, which states that higher return volatility is correlated with a greater reliance on non- interest income The same conclusions were reached by Stiroh and Rumble (2006), Jaffar et al.

(2014), Senyo et al (2015) who contend that heightened systemic risk is linked to the banks' increased dependence on non-interest income Besides, services that generate non- interest income are rather transactional, meaning that clients can quickly and without hesitation switch service providers Payment processing can be taken as an example - as this automated service can be performed by almost any bank in a similar manner, there is no competitive differentiation between banks other than competing on its price As there is very little that commercial banks can do to avoid customer churn, non-interest source of income may not be a preferable option for them.

According to Hughes et al (2003), diversifying banking services into non-core activities could make the structure and governance of commercial banks more complex and make it more difficult to effectively supervise banking activities As banks expand their non-interest related activities, information asymmetry and agency issues would become worse The study shows that although banks could also pursue their management infrastructure strategies, this might increase the costs and the complexity of the administrative systems, which reduces the efficiency of banks Hence, the influence of non-interest income on the performance of commercial banks may have constraints, and the interaction between interest income and non-interest income must be managed.

However, it is irrefutable that tremendous benefits have been brought to0 commercial banks by generating their revenue from non-interest income (Gallo & Kolari, 1996; Mabwe & Web, 2014; Saunders Schimid, 2014) First, the relatively stable population combined with the intense competition in the current age might cause limitations for commercial banks that merely concentrate on the traditional interest income In contrast, developing the non-interest income business can contribute to diversification and stabilization of the commercial bank's overall income even if the external environment is undergoing different business cycles Second, since interest income can fluctuate with interest rates, the income source of the banks might become quite erratic if they depend solely on this traditional income Contrary to interest income, non-interest income is typically not impacted by economic and financial market cycles and is not typically subject to regulation or legal restrictions Accordingly, they are less fluctuating than interest income Third, banks can save on fixed costs by facilitating the distribution of several business lines using customer information For instance, providing customers with information when making loans could also include offering them additional financial services like financial consulting or securities underwriting, and vice versa (Stein, 2002).

Raising the non-interest income ratio could also have the impact of encouraging the bank to devote more attention to developing its brand and image as well as its recruitment and training strategies As modern services are often new to the majority of customers as well as many potential customers who have not yet developed a need to use the service, it is required that each bank has its own strategy to develop an image to highlight its advantages, role and position to attract more customers Moreover, noninterest services also drive interbank collaboration such as transferring money or using payment cards to serve the needs of customers, thereby benefiting banks by reducing their operational costs and utilizing fixed expenses since they may share inputs like manpower and technology for a variety of operations (Kevin J Stiroh 2004).

In terms of competitive perspective, Landskroner et al (2005) and Lepetit (2008) agree that such income diversification will create competitive pressure among banks in broader market segments which leads to the promotion of technological advancement and1 the efficiency of banking services The theory of finance and income diversification implies banks that provide a wider range of goods and services would generate more revenue Evidently, the development of non-interest activities could boost the income from service fees such as check payment fees, money transfer fees, entrustment and consulting fees, account management fees, card service fees and other types of income.

Baele (2007) claims that instead of focusing solely on traditional interest-based services, banks can collect more information through the diversification of activities, which facilitates the cross-selling of products and the development of other activities. Considering the benefits of cross-selling products, Klein & Seidenberg (1997) and Mester

(2010) contend that banks gain economies of scope when diversifying their sources of income through non-interest banking activities With the advantage of serving a large number of customers, non-credit activities also play a complementary role and support other activities, thus indirectly generating profits from activities such as maintaining payment accounts for organizations through payroll services and payment card issuance.

It can be seen that notwithstanding lending activities, non-interest activities also provide safe and stable sources of income (Klein & Saidenberg, 2005) Boyd and Gertler

(1994) show that the movement toward off-balance sheet activities does not indicate that banks are abandoning their traditional lines of business In fact, the diversification into non-interest income is being pursued across banks for spreading risks and reducing dependence on the main source of income (Boyd & Gertler 1994; Hoang Ngoc Tuyen &

Vo Thi Hien, 2010) According to the Modern Portfolio Theory developed by Markowitz in 1952, banks can maximize expected returns to reduce risk or return volatility by diversifying income to hedge against market risk and concentration risk This theory states that whether or not a bank can diversify depends on the factors affecting credit risk and the bank can eliminate credit risks caused by macro factors through the governance structure The theory also suggests that non-interest income catalyzes the reduction of bank’s credit risk (Hunjra et al., 2020).

The development of non-interest banking services is an evitable trend in the

Vietnamese economy to meet the increasing requirements of customers, especially in the2 context of international integration and development Undoubtedly, non-interest income is an essential source of revenue for commercial banking organizations Despite the fact that accelerating the transformation of banking services through non-interest operations may have a number of advantages, there is generally a very high cost associated with it as well.Therefore, commercial banks need to carefully weigh the pros and cons before they finally make a decision.

Empirical review

Numerous studies have analyzed the association between non-interest income and bank performance to determine whether this type of income aids in the proliferation of commercial banks There are two strands of literature on the effect of non-interest income on the performance of commercial banks On the one hand, banks with greater diversification through non-traditional sources of income are capable of reducing income volatility substantially as well as improving their performance (Smith et al., 2003; Baele et al., 2007; Lepitit et al., 2008) Meslier (2014) hypothesizes that a shift towards non- interest income will drive bank profitability and risk-adjusted profits This means banks that are able to diversify their income by concentrating on non-interest income streams could catalyze their performance and even stability (Odesanmi & Wolfe, 2007; Kohler, 2014; Senyo et al., 2015; Rayenda, 2018) These results are in line with the results of Sheng & Wang (2008) on the Chinese banking industry, Pennathur et al (2012) on Indian banks, and Aslam et al (2015) on Pakistan banks.

Applying the Data Envelopment Analysis, Elyasiani and Wang (2012) examine whether non-interest income is associated with an improvement or reduction in the efficiency of banks The study concludes with three main channels explaining how non- interest income could facilitate bank performance First, with the rising concentration on non-interest activities in today’s market, banks have to face the increasing competition between banks, finance companies and insurance firms Therefore, banks are forced to improve their efficiency in management and maintain competitive advantages through the advancement of science and technology, which could boost bank performance effectively.

Second, when banks diversify their sources of income through non-interest banking3 activities, they could gain economies of scale by cross-selling and the reuse of inputs. Compared to banks that focused on traditional activities, these benefits enabled banks to maintain competitive advantages and increase revenue (Hughes et al., 2001; Stiroh, 2004). Third, the sharing of customer information of non-interest income characteristics can enhance the bank performance In particular, the financial sector is highly information- intensive (Diamond, 1984; Denis & Mihov, 2003), but information or data about the bank customers could be shared with subsidiaries without additional costs (Saunders & Walter, 1994; Kashyap et al, 2002).

On the other hand, there are studies stipulating that the impact of non-interest income on commercial banks is limited or even inversely correlated with their performance As banks convert from traditional activity focus to non-interest related services, the income volatility exacerbates (De Young & Roland, 2001) Thus, banks that expand their reach into non-traditional activities have a higher level of risk than banks that concentrate solely on traditional activities (Lepetit et al., 2008; Williams, 2016) and through the years the volatility of non-interest revenue would grow (Rosie, Christos & Geoffrey, 2003; Hirtle & Stiroh, 2007) Studies by DeYoung and Roland (2001) conclude that there are three key factors that account for the risk associated with the performance of commercial banks induced by non-interest income First, compared to traditional banking activities, transitioning to non-interest ones is more affordable Consequently, non-interest income is highly volatile and may have a detrimental impact on the performance of commercial banks Second, the expansion of non-interest activities might force banks to increase their fixed costs Therefore, the bank’s initial cost of non-interest services is rather higher than that of utilizing interest-based banking activities These circumstances could escalate the operating leverage of banks, thereby potentially raising their exposure to risk According to Trivedi (2015), the research concludes that while increased non- interest income share has a favorable impact on profitability, it does not have risk-adjusted impacts, and as a result, does not guarantee the stability of returns and the bank performance These results are also consistent with the study of Stiroh (2004) on U.S banks, Melister (2014) on Mexican banks and Sun et al (2017) on Chinese banks.4

There are empirical literature on the effect of non-interest income on bank performance that are mixed Based on a sample of 10341 banks in the U.S from 2002 to

2013, the research result of Saunders et al (2014) shows that a higher ratio of non- interest income to interest income is associated with a higher performance across the banking sector, while Chiorazzo et al (2008) find this diversification effect to be stronger for large banks and the source of these non-traditional services to be less important Meanwhile, DeYoung & Torna (2013) find that the likelihood of bank failure tends to increase in certain non-traditional business sectors like investment banking, venture capital and asset securitization Contrarily, non-traditional activities like insurance sales and securities brokerage may reduce the probability of bank failure In addition, examining Indian banks based on data for the period 1998 - 2014, Ahamed (2017) finds that greater diversification of revenue in the form of non-interest income enhances bank profits and risk-adjusted profits Yet, banks that had lower asset quality, in terms of the share of non-performing loans or loan loss provisions, can reap higher income diversification benefits compared to the banks that had higher asset quality.

Many existing studies on emerging market economies yield similar evidence. Using a sample from 11 emerging countries from 2000 to 2007, Sanya and Wolfe (2010) show that a higher ratio of non-interest income in the bank revenue is significantly associated with lower bankruptcy risk, higher profitability and market value Meanwhile, examining banks from 12 Asian economies between 2001 and 2015, Salike and Ao (2018) suggest that while non-interest activities could increase profitability or reduce risks for banks in middle and low-income countries, they might raise risks for banks in high- income countries On the contrary, Lee et al (2014) examine 22 countries in Asia over the period 1995 - 2009 and explore that non-interest activities of Asian banks reduce risk, but do not increase profitability Moreover, regarding the importance of non-interest income in affecting the performance of Asian banks, Phan Thi Thu Hang (2023) claims that the study results in developing nations are distinct from those in developed nations Notably,the performance of commercial banks in the ASEAN region is negatively impacted by non-interest income However, this study suggests that diversification is essential in the5 current environment of banking competitiveness Hence, commercial banks should recognize the latest market demands, offer products and services that are in accordance with the most recent trends, and refrain from offering an excessive number of utility items to significantly avoid fluctuations in non-interest income.

There are articles that produce new evidence for the experimental results on the benefits of non-interest income during the crisis period While investigating how noninterest income influences the risk and return of banks in the U.S between 2007 and

2009, Park (2019) finds that non-interest income could reduce bank risk and increase the stock return in times of financial crisis Similarly, using a sample of 30 commercial banks operating in Kenya from 2010 to 2020, Ochenge (2022) then reports that reliance on noninterest revenue sources could act as an economically important shock absorber for banks in times of declining profits like during the Covid-19 outbreak In the studies of Govori (2013), Vithyea (2014) and Senyo et al (2015), the results show that banks can establish a buffer against credit risk and manage to boost liquidity by growing their non- interest income, identified in the cases of Cambodia, Kosovo, and Ghana respectively. Nevertheless, when Sherika and Ajit (2022) employ data between 1996 and 2018 for 24 Asian countries to examine this relationship, their results show that despite differences among the country characteristics, it is apparent that a focus on non-interest income does not render banks more stable, especially in the post global financial crisis period Saunders et al (2014) also find no convincing evidence that noninterest- generating activities harm bank performance or increase bank failure, insolvency, or systematic risks during both crisis and noncrisis periods.

In the context of Vietnam, the empirical evidence on the impact of non-interest income on the performance of commercial banks also provides mixed results Many studies have analyzed the reciprocal association between non-interest income and bank performance In fact, not only can the expansion of non-interest activities help Vietnamese commercial banks increase their performance, but it is also an inevitable development trend for Vietnamese commercial banks in the current integration and competition economy (Ho Thi Hong Minh & Nguyen Thi Canh, 2015; Le Long Hau & Pham Xuan6 Quynh, 2016; Nguyen Minh Sang & Tran Thi Thanh Tam, 2020).

Using data from 22 Vietnamese commercial banks between 2013 and 2022, Nguyen Quoc Anh and Tang My Sang (2022) provide additional insights into the effects of non-interest income and explore that non-traditional banking services may have hidden risks, but if commercial banks could take advantage of such opportunities and have appropriate business plans, these activities could bring about increased revenue for the banks Next, when investigating the impact of income diversification on bank risk in Vietnam before and during the Covid-19 pandemic, Thanh Tam Le (2022) argues that the diversification strategy of banks should be conducted based on the sources of non-interest income, as the higher the level of income diversification, the lower the default risk.

However, the detrimental effect of non-interest income on the performance of banks has been the subject of numerous previous studies in Vietnam Conducting research from 21 Vietnamese commercial banks during 2007 and 2017, Cuong Van Hoang (2020) draws the conclusion that Vietnamese commercial banks should remain focused on their traditional interest-based activities rather than diversifying towards non- interest services since this may prompt both a lower level of profitability and liquidity creation Besides, expansion into non-traditional activities does not guarantee superior performance (Nguyen Khanh Ngoc, 2019) The regression results indicate a negative correlation between noninterest revenue and the efficiency of Vietnamese commercial banks.

In addition, using the GMM estimator to test a sample of 13 Vietnamese listed banks between 2010 and 2019, Le Hai Trung (2021) further notes the fact that although income diversification has a negative impact on bank performance, it does not necessarily imply Vietnamese commercial banks should avoid non-interest activities This association may be influenced by recent significant bank investments in the infrastructure of core banking, leading to a short-term reduction in the profit margin Therefore, this finding implies that Vietnamese commercial banks should increase their operational flexibility and efficiency in order to take advantage of the quick changes in the business environment brought on by the fourth industrial revolution's advancements in financial technology.

Also, the explanation of prior research in Vietnam is that either unique factors associated7 with the Covid -19 crisis or the effect of the recent development of Vietnamese banking infrastructure triggered the unbeneficial impacts of non-interest income on the performance of banks.

DATA AND METHODOLOGY

Estimation procedure

There are six stages that this thesis uses to address the research questions:

To serve the research, the author uses the method of collecting secondary data by taking the published data from the annual reports of Vietnamese commercial banks from

2011 to 2021 at Vietstock and Cafef websites as well as the data from the websites of commercial banks under study, websites of the World Bank and the General Statistics Office of Vietnam.

This thesis then conducts the methodological review to find out the determinants applicable to the dataset Next, by using Stata 17 s t atistical software, a summary description of the data characteristics of the dependent variable and the independent variables such as mean, maximum, minimum and standard deviation is performed.

Step 3 Analyze the correlation matrix: The correlation matrix is done to analyze and check the correlation between the research variables Subsequently, the article implements the Variance Inflation Factor (VIF) to detect multicollinearity.

Step 4 Test the model by Pooled OLS, FEM, REM:

The regression methods, including the Pooled OLS, FEM and REM, are used to generate appropriate results based on time series and cross-sectional data The author then conducts several tests including the F-test and Hausman test to discover the best- fitted regression model for this study.

Step 5 Check the model defects: In order to increase the reliability and relevance of the research results, the thesis tests the model defects, including multicollinearity, autocorrelation and heteroskedasticity Thus, the author needs to apply the FGLS method to eliminate any problems of the research model that occur To ensure the most appropriate results, the author uses the GMM estimator to handle the endogeneity problem Finally, Independent Sample T-Test would be used to check the robustness of the models.

Step 6 Conclusion and policy implications: Based on the results of the regressions, the author draws conclusions, and makes recommendations to improve the bank performance.

Data collection

The data used in the study were collected from the audited financial statements and annual reports of 26 Vietnamese commercial banks (out of the possible 31 Vietnamese commercial banks) throughout an 11 year-period, between 2011 and 2021 Due to the influence of the global economic crisis as well as the Covid-19 outbreak and recovery, the banking industry has faced numerous ups and downs over this period, which would demonstrate the effect of non-interest revenue on the performance of commercial banks more efficiently Because of limited research time and incomplete data sources, the researcher selected banks that own complete financial statements, including balance sheets,income statements, cash flow statements, and financial statement notes In addition,information was gathered from the websites http://finance.vietstock.vn and https://cafef.vn/,websites of commercial banks under study, websites of the World Bank and the General Statistics Office of Vietnam The data were then imported into an Excel file to be edited and encoded The following stage was to conduct data cleaning to detect errors including blank cells with lack or incorrect information and complete the data matrix Finally, using Stata 17 software, the data were processed and calculated.

Table 3.1 The lists of selected Vietnamese commercial banks

1 An Binh Commercial Joint Stock Bank ABB

2 Asia Commercial Joint Stock Bank ACB

3 Bac A Commercial Joint Stock Bank BAB

4 Joint Stock Commercial Bank for Investment and Development of Vietnam BID

5 Viet Capital Commercial Joint Stock Bank BVB

6 Vietnam Joint Stock Commercial Bank for Industry and Trade CTG

7 Vietnam Commercial Joint Stock Export Import Bank EIB

8 Ho Chi Minh City Development Joint Stock Commercial Bank HDB

9 Kien Long Commercial Joint Stock Bank KLB

10 Lien Viet Post Joint Stock Commercial Bank LPB

11 Military Commercial Joint Stock Bank MBB

12 Vietnam Maritime Commercial Joint Stock Bank MSB

13 Nam A Commercial Joint Stock Bank NAB

14 National Citizen Commercial Joint Stock Bank NVB

15 Orient Commercial Joint Stock Bank OCB

16 Petrolimex Group Commercial Joint Stock Bank PGB

17 Saigon Bank For Industry And Trade SGB

18 Saigon Hanoi Commercial Joint Stock Bank SHB

19 Southeast Asia Commercial Joint Stock Bank SSB

20 Sai Gon Thuong Tin Commercial Joint Stock Bank STB

21 Vietnam Technological and Commercial Joint Stock Bank TCB

22 Tien Phong Commercial Joint Stock Bank TPB

23 Vietnam - Asia Commercial Joint Stock Bank VAB

24 Bank for Foreign Trade of Vietnam VCB

25 Vietnam International Commercial Joint Stock Bank VIB

26 Vietnam Prosperity Joint Stock Commercial Bank VPB

Source: The author’s own summarization

Research methodology

In line with prior studies (Chiorazzo et al., 2008; Ho Thi Hong Minh & Nguyen Thi Canh, 2015; Nguyen Minh Sang, 2017; Le Long Hau & Pham Xuan Quynh, 2017), this study employs regressions with dependent variables are ROA, ROE; and noninterest income is independent variable, whereas bank size, asset quality, bank capitalization, loans to assets ratio, deposits to assets ratio, total costs to assets ratio, loan loss provision ratio, the growth rate of gross domestic product, the rate of inflation are control variables Therefore, this thesis examines the relationship between non-interest income and bank performance by estimating the following model:

Performance (ROA, ROE) i,t = α + β1 NNII i,t + ∑ s=2ps λ i,t + + ε i,t

In which i represents the number of banks in the research sample, i = 1 … 26; t represents the time (t = 2011 - 2021), β represents the regression coefficient, λ is the control variable matrix, ε is the random error Measuring efficiency by ROA, ROE; NNII represents the ratio of net non-interest income variable Control variables include bank size (SIZE), asset quality (NPL), bank capitalization (CAP), loans to assets ratio (LOAN), deposits to assets ratio (DEP), total costs to assets ratio (COST), loan loss provision ratio (LLP), the growth rate of GDP (GDP), the rate of inflation (INF).

(1) ROA i,t = α + β 1 NNII i,t + β 2 SIZE i,t + β 3 NPL i,t + β 4 CAP i,t + β 5 LOAN i,t + β 6 DEP i,t + β

(2) ROE i,t = α + β 1 NNII i,t + β 1 SIZE i,t + β 2 SIZE i,t + β 3 NPL i,t + β 4 CAP i,t + β 5 LOAN i,t + β

6DEP i,t + β 7 COST i,t + β 8 LLP i,t + β 9 GDP t + β 10 INF t + ε i,t

Followed is the figure depicting the conceptual framework of the research model:

Bank size Non-interest income

Deposits to assets The annual growth rate of GDP

The rate of annual inflation

Figure 3.1 The conceptual framework of the research model

Total costs to total assetsLoan loss provision

Following the previous studies of Chiorazzo et al., (2008), Ho Thi Hong Minh and Nguyen Thi Canh (2015), this study uses return on assets and return on equity as indicators to measure the bank performance The formations of given variables are computed as follows:

Average totalassets + Return on equity:

Non-interest income: Based primarily on the theoretical models of Clark Siems

(2002), Deyoung & Rice, (2004), Valverde and Fernandez (2007), Nguyen (2012), the thesis uses non-interest income (NNII) measured based on the ratio of non-interest income to total assets to analyze the impact of non-interest income on the performance of commercial banks:

Non-interest income NNII Total assets

This ratio is annualized income from bank services and sources other than interest-bearing assets, divided by total assets This ratio reflects the average assets of a commercial bank that will generate non-interest income over a course of time This approach can be used to make predictions about future development trends and the overall contribution of non-interest income to the bank performance (Nguyen MinhSang & Nguyen Thi Hanh Hoa, 2012) Some studies calculate the non-interest income ratio based on non-interest income divided by the sum of net interest income and

4 noninterest income (Chiorazzo et al, 2008; Elsas et al, 2010; Ho Thi Hong Minh & Nguyen Thi Canh, 2015; Le Long Hau & Pham Xuan Quynh, 2017) However, this methodology only accurately captures the meaning of the ratio of non-interest income when the bank's non-interest activities and credit activities both have positive values, otherwise, an increase or decrease in this ratio cannot fully reflect the trend of non- interest income Therefore, non-interest income to total assets is highly recommended to measure the ratio of non-interest income (Hoang Ngoc Yen & Vo Thi Hien, 2010).

Most research conducted outside of the United States that examined the influence of diversification on banking performance came to the conclusion that non- interest revenue has beneficial effects on banks (Ochenge, 2022) According to Chiorazzo et al (2008), non-interest revenue and risk-adjusted earnings for the Italian banking sector are positively correlated For large banks in particular, the influence of non-interest income is seen to be stronger, albeit the source of this money is discovered to be less significant Meslier et al (2014) indicate that non-interest activities have a positive impact on the performance of Philippine banks Subsequently, when Nisar et al.

(2018) analyze a large panel of banks from South Asian countries, they find that a shift to non-interest income has a favorable impact on risk-adjusted profits Ammar and Boughrara (2019) also find a reciprocal effect of non-interest income for a sample of banks from the Middle East and North Africa region.

H1: Non-interest income exerts positive impacts on Vietnamese banking performance

Bank size: Natural logarithm of total assets is included as an independent variable to measure bank size (Nguyen, 2012) Larger banks are supposed to be less risky since they may benefit from a more diverse investment portfolio and sophisticated risk management Nonetheless, larger banks could reveal a greater exposure to risk due to their business models, which usually feature a greater proportion of higher-risk investment banking operations, thus these banks may have to suffer lower returns(Athanasoglou, Delis & Staikouras, 2006) In addition, larger banks typically have more

5 extensive administrative structures, partially as a result of greater regulation, and diseconomies of scale, which could lead to lower profitability Large banking institutions, according to Vallascas and Keasey (2012), are more inclined to make riskier investments, which might therefore perform less effective than small banks. However, Salike and Ao (2018) claim that bank size can avail of the advantages of economies of scale as well as provide protection from negative shocks during recessions At the same time, a bank’s profit growth is seen as more stable when its assets expand faster (Nguyen Khanh Ngoc, 2019) In addition, the "too big to fail" theory stipulates that large financial companies should not be allowed to fail due to the potential adverse impact their failure may pose on the rest of the sector and the economy at large Therefore, these banks are usually supported by the government Hence, the larger the size of a bank’s assets, the more likely it is to increase security and revenue, expand its assets, as well as use capital more efficiently (Molyneux & Thornton, 1992; Bikker & Hu, 2002 and Goddard et al., 2004).

SIZE = Natural logarithm (Total assets)

H2: Bank size exerts positive/negative impacts on Vietnamese banking performance

Asset quality: The ratio of non-performing loans to total loans represents the quality of assets (Salike & Ao, 2018) Asset quality depends primarily on its loan portfolio and internal credit management system Poor asset quality, commonly referred to as poor loan quality and normally represented by non-performing loans or impaired loans, is an important consideration in asset administration and an indicator of potential banking performance Non-performing loans are loans that have been outstanding for at least 90 days and have been prolonged for a considerable amount of time without generating income Most non-performing loans typically relate to bank collapses and financial crises in both emerging and industrialized nations (IMF, 2009).The excessively high level of non-performing loans may indicate lax credit administration processes or a lack of credit risk management procedures Research byMester (1996), Berger and DeYoung (1997), Berger and Mester (1997), Nathan et al.

(2020) states that the effect of NPL can be devastating to an economy if not controlled and this would have a negative implication on the whole banking industry.

Non-performing loans Total loans

H3: Asset quality exerts negative impacts on Vietnamese banking performance

Bank capitalization: The equity to total assets ratio is an indicator used to measure financial leverage Most recent research also use this variable such as the studies of Stiroh(2004b), Chiorazzo et al (2008) and Sanya & Wolfe (2011) It is expected that the higher the equity-to-asset ratio, the lower the need for external funding, and therefore, the higher the bank performance In addition, the higher the ratio of equity to assets, the smaller the risk of insolvency that banks might face (Kosmidou et al., 2008; Caporale et al., 2017; Mirzaei et al., 2013) Experimental evidence by Berger (1995), Kunt and Huizinga (1999) and Goddard et al (2004) suggests that the ideal banking activities are to keep a high level of equity corresponding to bank assets, indicating bank capitalization exerts favorable impacts on the bank performance.

H4: Bank capitalization exerts positive impacts on Vietnamese banking performance

Loans to assets: The ratio of total loans to total assets is used to control for variations in the asset composition (Stiroh & Rumble, 2006; Chiorazzo et al., 2008; Chortareas et al., 2011; Meslier et al., 2014) It captures the bank's involvement in market-based activities (Sanya & Wolfe, 2011) The higher the loan ratio, the more aggressive a bank should be toward profitability because it has a substantially greater portion of interest-bearing assets (Claeys & Vander, 2008; Hesse & Poghosyan, 2009).

It is undeniable that the primary income source of a bank comes from lending activity, which is expected to have a positive impact on the bank performance However, if a bank has a large loans-to-assets ratio, it may negatively affect its performance as the

NPL 7 bank’s risk would rise According to research by Chiorazzo et al (2008), Ho Thi Hong Minh & Nguyen Thi Canh (2015), Le Long Hau & Pham Xuan Quynh (2017), the ratio of loans to total assets has a beneficial impact on the performance of commercial banks.

H5: Loans to assets ratio exerts positive impacts on Vietnamese banking performance

RESEARCH RESULTS AND DISCUSSIONS

Descriptive statistics

This study aims to assess the impact of non-interest income on the performance of Vietnamese commercial banks The panel data is unbalanced as several bank- specific variables are missing from the data sample, with the lowest observation being NPL, at

285 bank-time observations Table 4.1 briefly summarizes the informational coefficients in the data set through mean, standard deviation, min, and max.

Table 4.1 Descriptive statistics Variables Observation Mean Standard Deviation Min Max

Source: The author’s own estimation

Statistical results in table 4.1 show that the average ROA of all banks over the sample period is in the vicinity of 1.2 percent, whereas the value of the ROE ratio is 10.15 percent on average Meanwhile, the performance varies notably between banks as the standard deviation of ROA and ROE are 2.07 percent and 8.4 percent, respectively.

Figure 4.1 gives comparisons between ROA and ROE ratios of the bank data set over an 11-year period, beginning from the year of 2011.

Figure 4 1 Return on assets and return on equity ratios

ROE ROA Source: The author’s own estimation

Between 2011 and 2021, the percentage of ROA and ROE experienced an upward trend, while ROA climbed from 1.19 percent to 1.23 percent, ROE surged 0.46 percent from 9.8 percent in 2011 and reached the mark of 10.26 percent at the end of the period.

Particularly, from a negligible of 1.19 percent in 2011, the percentage of ROA showed a small and gradual decline in the four following years In fact, it was not until

2016 that the proportion of ROA shot back up again and accounted for 1.2 percent. From then on, this figure rose remarkably before reaching a peak of 1.23 percent in

2021 However, such discrepancies could not be seen in the ratio of ROE throughout the same timeframe By 2021, ROE ratio soared to north of 10.2 percent, after showing a continuous and steady increase in the period of 2011-2016.

It is apparent that ROA and ROE ratios both registered persistent growth despite a slight fall in the allocation of the banks’ ROA in five initial years The explanation for threatened to be destroyed by the severe worldwide economic crisis, called “The Great Recession” In Vietnam, commercial banks were also negatively affected by the bad debt circumstance, which incurred higher expenses in provisioning, administrative and operating activities Consequently, Vietnamese commercial banks were compelled to undergo a recovery process and system reorganization between 2011 and 2015. Evidently, Vietnam had enacted Decision No 254/QD-TTg dated March 01, 2012 of the Prime Minister on "Restructuring the system of credit institutions during the period of

2011 -2015”, facilitating the latter escalating in the following years.

The results of the statistical analysis described in table 4.1 show the non-interest income ratio of Vietnamese commercial banks has an average value of 0.6 percent in the period 2011-2021 Among 26 Vietnamese commercial banks, the lowest and the highest NNII ratios are -0.006 (ACB) in 2012 and 0.0275 (TCB) in 2017, correspondingly.

Figure 4.2 Non-interest income ratio

Source: The author’s own estimation

In figure 4.2, it is evident that the non- neres ncome rao wnesse a rising trend between the year of 2011 and 2021 A slight volatility in the growth period of NNII can be seen from 2011 to 2017, which hovered around 0.63 percent After that, this figure began to rapidly rocket to 0.64 percent at the end of the period This indicates that Vietnamese commercial banks have started to have achievement in improving the bank performance after the implementation of Decision No 254/QD- TTg of the Prime Minister on reforming the banking system throughout 2011-2015, along with the goal of decreasing the bank’s reliance on credit activities and increasing their income from noninterest services of Decision No 986/QD-TTg of the Prime Minister on “Approving the Strategy for development of Vietnam’s banking sector through 2025, with orientations toward 2030” Therefore, it is undoubted that the growth direction of domestic commercial banks is toward expanding non-interest earning operations.

The results in table 4.1 show that with a standard deviation of 1.14, the average bank size is approximately 32.43 The largest bank size is 35.11 (BID) in 2021 while the smallest is merely 30.32 (SGB) in 2013.

The loan to assets ratio (LOAN) is 56.31 percent average This ratio varies substantially between banks with the highest and lowest LOAN which are 14.48 percent (TPB) in 2011 and 78.81 percent (BID) in 2020 respectively, suggesting that almost half of the assets held by most banks are used for lending activities In figure 4.3, despite a states that the orientation of commercial banks is to boost the revenue from non-interest activities in tandem with maintaining a controlled growth of interest activities since credit services are still the most lucrative services for banks in the present time.

Deposits accounts for 65.49 percent on average of the total assets across banks. With a standard deviation of 11.99 percent, the minimum and maximum value of DEP ratio are 25.1 percent (TPB) in 2011 and 89.4 percent (STB) in 2015 This means that customer deposits remain the core source to generate the bank income.

The mean of NPL and LLP are all less than 5 percent, which are 2.19 percent and 4.35 percent, sequentially To be specific, the bank that has the smallest NPL ratio belongs to TPB in 2013 (0.35 percent), whereas the NPL ratio of CTG in 2015 is the highest (9.18 percent) Meanwhile, it is VIB and SGB that has the lowest and largest LLP ratio, which are 0.001 percent in 2021 and 826.8 percent in 2016, correspondingly. Besides, CAP ratio has the mean value at 9.13 percent and the standard deviation of 3.77 percent, revealing that the majority of banks use hedging strategies to reduce the risk of bad debt The analysis results show that the bank has the lowest capitalization is BID at 4.06 percent in 2017, while the largest is SGB at 23.84 percent in 2013.

In accordance with figure 4.3, there was a noticeable difference in the fluctuations of NPL, CAP and LLP ratios to that of COST ratio The ratios of NPL,CAP and LLP witnessed downward trends between 2011 and 2015, before showing maintains its rate of increase during the whole period This ratio has an average value of 1.64 percent and a standard deviation of 0.6 percent, with TPB having the largest cost to assets value at 5.19 percent in 2011 and SSB registering the lowest value, at 0.097 percent in 2014.

In terms of macroeconomic variables, between 2011 and 2021, the mean value ofGDP and INF are 5.9 percent and 5.15 percent, respectively Particularly, GDP which presents the growth of the economy has a standard deviation of 1.6 percent, in association with the minimum value which is 2.6 percent in 2021 and the maximum value which is 7.5 percent in 2018 Meanwhile, the standard deviation of INF ratio is4.79 percent, and the year that has the largest inflation rate is 2011 (8.58 percent) and the lowest is in 2015 (0.6 percent) As shown in figure 4.3, the INF ratio began to decline after reaching its peak in 2011, indicating a promising situation for theVietnamese economy.

Source: The author’s own estimation

Source: The author’s own estimation

Figure 4.3 The fluctuation trends of control variables between 2011 and 2021

Correlation matrix

Table 4.2 Correlation matrix and multicollinearity

ROA ROE NNII SIZE NPL CAP LOAN DEP COST LLP GDP INF VIF

Source: The author’s own estimation

In table 4.2, multicollinearity and correlation matrix were conducted as part of preliminary assessment of variables.

First, it is said that if the correlation exceeds 0.8 then severe multicollinearity may be presented (Gujarati, 2004) The analysis result in table 4.2 demonstrates that there is no absolute value of the correlation coefficients between the research variables that are higher than 0.8, with the highest value being the correlation between CAP and SIZE, at 0.5755 This shows that indicators in the model have a relatively small correlation Hence, it can be concluded that the likelihood of multicollinearity in the regression models between the independent variables is not significant.

As shown in the table, NNII and LOAN are beneficially related with ROA and ROE, whereas variables such as LLP, GDP and INF pose an unfavorable interrelationship with both bank performance indicators In addition, there is a similarity between the directions of CAP, DEP and COST that can also be witnessed, which positively affects ROA while having inverse correlation with ROE Conversely, although SIZE adversely links to ROA, it has a reciprocal association with ROE.

Second, multicollinearity occurs when an explanatory variable is highly correlated with a linear combination of the other independent variables, which will trigger less trustworthy statistical inferences In other words, the existence of multicollinearity in a data set can lead to less reliable results due to larger standard errors This multicollinearity issue is measured by the Variance Inflation Factor (VIF).

If the result of the VIF test shows that all coefficients are less than 10, multicollinearity does not occur in the investigated number (Kennedy, 1992) The VIFs of all variables in table 4.2 are all down to satisfactory values, which are all less than 3 This means that the multicollinearity does not seriously interfere with the estimated models.

Empirical results

Table 4.3 Results of Panel Regressions

Estimations Pooled OLS FEM REM FLGS SYS-GMM

Variables ROA ROE ROA ROE ROA ROE ROA ROE ROA ROE

1 Denotes significance at 10% level Source: The author’s own estimation

Estimations Pooled OLS FEM REM FLGS SYS-GMM

Variables ROA ROE ROA ROE ROA ROE ROA ROE ROA ROE

Table 4.3 gives information about regression findings on bank performance measured by return on assets and return on equity While columns 1-2, columns 3-4 and columns 5-6 report the results from Pooled OLS, fixed and random effects panel data estimators respectively, the results from the Feasible Generalized Least Squares and System Generalized Method of Moments are reported in columns 7-8 and columns 9-10 correspondingly It is evident that these methodologies depict contradicting empirical results between bank performance indicators Noticeably, non-interest income ratio is statistically insignificant with return on assets in Pooled OLS, FEM and REM, yet it has statistical meaning with the rest of measurements in both dependent indicators.

Regarding return on assets model, the results of the Pooled OLS approach in table 4.3 show that except for CAP and NPL ascertained at 5% and 10% significant level respectively, all other variables, including NNII which is the most important independent indicator, do not meet statistical thresholds Similar evidence could also be seen in the metrics of fixed effects regression and random effects GLS regression It is apparent that in both methodologies, SIZE, CAP, LOAN, COST, and INF all have statistical significance with ROA, whereas variables such as NPL, DEP, LLP, GDP, particularly NNII, are not.

The regression analysis of return on equity presented in table 4.3, meanwhile, highlights that NNII is significantly and positively correlated with return on equity across three measures, namely Pooled OLS, FEM and REM With the magnitude of 4.12 and the value of 1%, the estimations in Pooled OLS conclude that increased share of noninterest income accompanies beneficial effects with return on equity Simultaneously, the coefficients of NNII in FEM and REM are at the significance levels of 5% and 1%, correspondingly The empirical results in these metrics suggest that a unit change in non- interest income leads to 2.46 unit rise of return on equity

(FEM) or 3.72 unit rise of return on equity (REM) This means that there is a favorable correlation between the NNII and ROE, implying the focus on non-interest income would facilitate the bank performance As to other controlled variables, there is a positive relationship between the SIZE, LOAN, INF ratio and ROE in all three estimation methods In Pooled OLS, opposite directions can be witnessed between NPL, DEP ratio and ROE whereas CAP, COST, LLP and GDP are statistically insignificant According to FEM and REM, COST adversely influences ROE In addition, although CAP and ROE are positively correlated in FEM, they do not have statistical meanings in REM. Meanwhile, NPL retains a statistically negative coefficient with ROE in REM, while it is not statistically significant in FEM.

Table 4.3 presents that the p-values of the F-statistics are less than 5% in both ROA and ROE models Therefore, the null hypothesis H0, which is Pooled OLS is more suitable than FEM, is rejected FEM is chosen as a result Subsequently, the Hausman tests were carried out to select between FEM and REM Regarding return on assets, the Hausman test has probability higher than 5%, which means REM is found to be the appropriate model Given the Hausman test of return on equity, the p-value is less than 5% indicating that using the FEM approach in the ROE model will produce a more consistent outcome than REM.

However, it can be seen in table 4.3 that while independent variables were able to explain changes in return on equity in the range of 42% to 47% in three initial models, R- squared of return on assets are merely 8.7%, 0.5% and 1.33 % in Pooled OLS, FEM andREM, correspondingly These values suggest that the percentage of the variance for a dependent variable which is explained by an independent variable is considerably low,which would result in unreliable final conclusions Furthermore, with the exception of multicollinearity, the results that are shown in table 4.4 reveal signs of heteroscedasticity and autocorrelation in all three conventional panel regressions This may render estimations from Pooled OLS, FEM, REM to become less accurate The detail results are reported in appendix 4 and appendix 5.

Table 4.4 Heteroscedasticity, autocorrelation and multicollinearity tests

Estimations Pooled OLS FEM REM

Dependent variables ROA ROE ROA ROE ROA

To guarantee a stable and effective estimate, Wooldridge (2002) claims that theFeasible Generalized Least Squares approach can be utilized to overcome the heteroscedasticity and autocorrelation phenomena Thus, instead of relying solely on previous conventional methods for regressions, FGLS is implemented Columns 7-8 of table 4.3 show that the model estimation results by the FGLS method are relatively compatible with the hypothesis The empirical result indicates a positive relationship between the non-interest income ratio and the performance of Vietnamese commercial banks with a regression coefficient of 0.29 compared with ROA model and 3.02 compared with ROE model, at the value of 1% This reciprocal relationship means that non-interest income benefits Vietnamese banks by enhancing their performance effectively For bank specific variables, findings of the ROA and ROE models both reflect that SIZE, CAP, INF are positively associated with bank performance, whereas NPL is outlined to reduce the performance of commercial banks In both models, other variables, namely LOAN, DEP, COST, LLP and GDP are statistically insignificant.

Next, by using System Generalized Method of Moments (SYS-GMM) estimation, according to Ahamed (2017), two significant econometric problems can be addressed: (i) SYS-GMM enables us to use lagged values of the dependent variable to exploit the data's dynamic nature; (ii) because the explanatory variables may not be strictly exogenous, endogeneity problems can be resolved while using lagged levels and lagged differences of the regressors as instruments Moreover, the System Generalized Method of Moments estimator is perceived to be suitable for small panel data with a large number of businesses and a small number of time series With regard to this study, the used panel data span only 11 years, while there are many banks involved, hence using the System Generalized Method of Moments approach will be appropriate.

To ensure the reliability of the model regression results, Arellano - Bond tests and Hansen tests are performed to verify the validity of the instruments and the potential error serial correlation First of all, table 4.3 shows that p-value of the Hansen test for ROA and ROE measures exceed 0.05, which implies that the instrument variables are valid and do not correlate with its errors, thereby the models are suitable So, neither model's variables are over-identifying Second, the results of Arellano - Bond tests show that the p-value of AR(2) of ROA and ROE models are greater than 0.05, thus indicating that models have no autocorrelation at the second-order difference So, the SYS-GMM estimator is consistent The research models used in this paper are highly sustainable.

The results of table 4.3 show that the majority of the coefficients are significant and have economically reasonable signs In particular, non-interest income has a positive interaction with the bank performance in the research phase at 1% significance level and a magnitude of 1.84 compared with return on assets and 10.12 compared with return on equity Regarding the results of control variables, the allocation of LOAN has a positive sign and is statistically significant with ROA at the 5% level In contrast, SIZE, COST, LLP, GDP ratio bear a negative relationship with ROA at the significant level of 1% (LPP), 5% (SIZE, COST, GDP) Also, NPL, CAP, DEP do not meet statistical meanings with ROA In the return on equity model, LOAN and COST have beneficial meanings with ROE whereas NPL, DEP, CAP, LLP, GDP and INF are seen to have negative impacts on ROE at the significant level of 1% (DEP, CAP, INF), 5% (NPL, GDP) and 10% (LLP) In addition, SIZE is not significant in ROE model.

Robustness Check: Independent Sample T-Test

To justify the findings, further research is conducted to test whether there is meaningful difference between banks with high NNII ratio and those with low NNII ratio.

If there are notable differences between these two groups, it can be contended that non- interest income does play a critical role on firm performance Before executing Independent Sample T-Test on the performance approaches, banks are first regrouped. Particularly, the data is two-part splitted, one part includes banks with high non-interest income ratios and the other are those with low non-interest income ratios.

ROA ROE diff t-statistics diff t-statistics

High-NNII-ratio banks and Low-NNII-ratio banks -0.004 -1.55 0.030 3.08 4

Source: The author’s own estimation

4 denotes significance at the 1% level confirmed regarding return on assets Nevertheless, the results indicate that there is a significant difference between high-NNII-ratio banks and low-NNII-ratio banks in terms of return on equity With the t-value corresponding to 1% level of significance, ROE of high-NNII-ratio banks is significantly different from that of low-NNII-ratio banks Therefore, these findings confirm the earlier result in which non-interest income plays important role on bank performance This result is in line with the results obtained by Chiarazzo et al (2008) and Meslier et al (2014).

Overall, the estimation results are generally comparable with the literature The findings estimated with the SYS-GMM are successful in resolving endogeneity bias and thus the SYS-GMM results are focused on the latter discussions First, the study observes that non-interest income ratio has statistical significance with return on assets and return on equity Second, the results confirm the existence of a strong relationship between noninterest income and bank performance since the magnitude of non-interest income ratio exerts a significant and positive impact on ROA and ROE This induces that banks which derive a larger share of their operating income from non-interest sources experience higher performance In other words, non-interest income will aid commercial banks in enhancing operational effectiveness, increased profitability, and ensuring security Third, the results find that non-interest income plays an important role on bank performance as high-NNII-ratio banks are experienced to have significant differences from low-NNII-ratio banks with regard to return on equity.

Result interpretation

4.4.1 The impacts of non-interest income on the performance of Vietnamese commercial banks

The regression results suggest a beneficial correlation between non-interest income and the performance of Vietnamese commercial banks The evidence reveals that non-interest income is a highly significant determinant of bank performance, such the performance of banks This is true for the research Hypothesis 1 given in Chapter 3. This conclusion is also reinforced by a number of earlier empirical studies by DeYoung and Rice (2004), Meslier (2014), Nguyen Minh Sang and Tran Thi Thanh Tam (2020).

Empirical research results have shown that developing in the direction of increasing the proportion of non-interest income is a solution for banks to improve their business efficiency Given the increasing competition on financial markets, Vietnamese banks are progressively shifting from traditional activities toward non- interest ones to maintain market share, reach out new consumers, and boost profitability In addition, there are several hazards associated with credit activities, as reflected by the high burden of bad debts in recent years, which have had a negative impact on the bank performance, prompting banks to focus on seeking profits from other non-credit activities Joint-stock commercial banks in Vietnam have thus expanded their sources of non-interest income to thrive and flourish in the context of international integration with fierce competition from domestic and foreign competitors.

Furthermore, in the mid-2021 and 2022, the prolonged Covid-19 pandemic has adversely affected the economy, causing the manufacturing and business activities to remain sluggish Meanwhile, the Vietnamese banking industry still retained bright colors in the profit picture during this stressed period In these years, fee income such as service fees, bancassurance fees was reported to be a vital growth engine of the Vietnamese banking sector and would continue to drive bank performance in the future. This positive result is in conformity with the studies of Park et al (2019), Ochenge

However, these conclusions deviate from the empirical evidence in the findings of Laeven and Levine (2007), Berger et al (2010) and Nguyen Khanh Ngoc (2019).Vietnamese commercial banks are still in the early phases of transitioning to offer a wide range of business lines, which means they lack experience in implementing and activities such as contributing capital, buying shares or trading securities Because the Vietnamese stock market is still in its infant stages as well as the legal framework governing these activities is still being developed and supplemented, it is currently challenging for Vietnamese banks to engage and earn high profit in such activities. Consequently, increasing involvement in non-traditional businesses might lead to lower operating efficiency of Vietnamese commercial banks Furthermore, it can be inferred that lending is still the most significant and traditional activity of Vietnamese banks. Therefore, the expansion of non-interest activities might eventually raise the overall risk to Vietnamese commercial banks.

4.4.2 The impacts of bank control variables on the performance of Vietnamese commercial banks

Estimation results show that for the case of Vietnamese commercial banks, size has no profound impact on return on equity, yet it exerts negative effects on return on assets This may imply that although economies of scale are crucial for performance, Vietnam financial markets do not always allow this advantage to facilitate higher performance Theoretically, a bank's capacity to withstand risk increases with its size (Lehar, 2005) However, this may not be consistent with the reality in Vietnam, where large banks have diminishing asset quality while the scale of credits and customers grows sharply There are also studies that do not detect a correlation between size and bank performance such as the research of Goddard (2004), Heffernan and Fu (2008), Mansour and Musaed (2021) This finding, however, contradicts to that of Goddard et al (2004), Salike and Ao (2018) which suggest that large scale banks can manage costs more efficiently, and thus leading to a better financial performance than smaller banks.Meanwhile, others see that as large banks typically have more extensive administrative structures, they are likely to suffer from bureaucracy which makes them slower in responding to economic changes, triggering lower performance (Vallascas & Keasey,

The research outcomes show that NPL has an inverse and significant impact on the ROE ratio whereas having no statistical significance with ROA, which suggests that banks should limit NPL ratio as a means of enhancing their efficiency In a high NPL scenario, banks usually have a growing propensity to tighten credit standards in response to deterioration in credit quality Besides, due to the high incidence of NPLs, banks generally raise their loan loss provision, which lowers their revenue and reduces the funds available for new lending This finding is in line with the studies of Mester

(1996), Berger and DeYoung (1997), Berger and Mester (1997).

The ratio of equity to total assets do not have statistical meanings with ROA. However, CAP ratio has a negative and statistically significant impact on ROE This could be explained by the fact that despite significant advancements in the Vietnamese banking sector, there are still shortcomings in the bank administration as well as difficulties that arise in many aspects, including the issue of equity Meanwhile, equity is extremely crucial in operating capital of commercial banks Therefore, Vietnamese commercial banks may easily be exposed to risks associated with income investment portfolio and off-balance sheet activities when supplementing capital This result is supported by Porter and Chiou (2013).

The research result of LOAN ratio is similar to those of Chiorazzo et al (2008),

Ho Thi Hong Minh and Nguyen Thi Canh (2015), Le Long Hau and Pham Xuan Quynh

(2017), suggesting that the ratio of loans to total assets is said to have a positive effect on the performance of commercial banks This result is consistent with the reality of Vietnamese commercial banks which prioritize lending to boost interest income.Although commercial banks are studying to expand non-interest activities to improve their performance, loans are still major sources of income generating assets for commercial banks As a result, when the bank's lending rate rises, the profit margin will also rise, thereby raising the bank's revenues. significantly associated with return on equity This result is in accordance with the results obtained by Le Long Hau and Pham Xuan Quynh (2017), Nguyen Minh Sang and Nguyen Thi Thuy Trang (2018) This is due to the fact that although the substantial number of client deposits may support the growth of credit operations, it is likely to increase the repayment pressure of banks which impedes the bank performance In other words, the higher the number of deposits in the banks, the higher expenses the banks have to pay for their customers such as deposit interest, thereby reducing the revenue of commercial banks This might serve as the foundation for banks to be aware of when structuring their operational capital.

Total cost to total assets has a negative regression coefficient with ROA and reaching 5% significance level This result corroborates those of Lepetit al al (2007),

Ho Thi Hong Minh and Nguyen Thi Canh (2015), Le Long Hau and Pham Xuan Quynh

(2017) Meanwhile, COST is shown to have a positive impact on ROE ratio at the significance level of 1% In fact, Vietnamese commercial banks have started to concentrate on non-interest activities to generate non-interest income, invest more in technological development, as well as focus on extending their networks Therefore, this would result in a rise in total cost to total assets of banks which could decrease the ROA ratio as the rising expenses yet increase ROE because of an enlargement in the activities of commercial banks.

Regarding loan loss provision ratio, the regression induces an inverse relation with both bank performance measurements This negative relationship shows that an increase in the LLP ratio reduces the performance of banks The finding implies that a percentage point increase in DEP may decline in overall 0.1 percent of the banks’ ROA and ROE, respectively This result vindicates the study of Larven and Majnono (2003),Mustafa et al (2012), Dietrich and Wanzenried (2011) The explanation for this is that because of the difficult economic situation in the banking industry in recent years, reducing their performance Also, a well-managed bank is stated to have a lower loan loss provision and such benefit would be translated into higher profitability (Ahmed, 2012).

It is shown in table 4.3 that GDP has an adverse effect on ROA and ROE at the meaning of 5% This can be explained that in Vietnam, domestic banks may face many challenges and anxieties with economic growth since the business environment is improved and barriers to entry financial markets are lowered in these circumstances, leading to higher competitions Furthermore, when GDP expands, it will contribute to an increase in the production of economic goods and services, forcing banks to develop new products to meet the requirement of customers and to strengthen their urgent solutions promptly in order to remain competitive in the marketplace In addition, as a rising GDP indicates that the economy is growing, the bank will use more assets for its primary activities As a result, the bank’s liquidity risk would increase and if the quality of the loan portfolio is poor, it will lead to an increase in loan loss provision Therefore, a negative relationship between GDP and bank performance can be expected Numerous studies reported similar results including the research of Tan and Floros (2012), Yanikkaya et al (2018).

The inflation rate has an adverse effect on ROE, whereas it is not confirmed to have statistical meaning with ROA, indicating a negative link between inflation and bank performance Inflation has an unbeneficial impact not only on the purchasing power and bank exchange rate, opportunity cost of retaining currency in the future, but it also could disrupt business strategies and negatively affect the performance of banks' equity holdings Undoubtedly, inflation might disrupt the bank’s business planning because of the uncertainties caused by the phenomena in both the pricing of services and the cost of inputs that reduced planned investment spending Similar evidence could also be witnessed in the research of Pervan et al (2015) and Mbabazize et al (2020).

This chapter has presented descriptive statistics of variables in the research model and highlights the fact that non-interest income is witnessing a rising trend in the Vietnam banking system In addition, correlation and multicollinearity are found neither in theROA nor ROE model The empirical findings of the study models have emphasized the favorable influence of non-interest income on bank performance and the relationship between control variables and the performance of Vietnamese banks using five regression techniques, namely Pooled OLS, FEM, REM, FLGS, and GMM.

CONCLUSIONS AND IMPLICATIONS

Conclusion

Due to the increasingly competitive banking market, which requires ongoing innovation to maintain the bank’s sustainability, the diversity of banking operations has become a topic of interest to the management of banking institutions, regulators, bank customers, and other stakeholders in recent years With the support of Stata 17 software, this research adds empirical evidence on the relationship between non- interest income and bank performance Specifically, five panel regression models including Pooled OLS, FEM, REM, FLGS and GMM are used with unbalanced panel dataset of 26 listed Vietnamese commercial banks, over the period from 2011 to 2021, to study the impact of non-interest income on bank performance.

The expansion of the bank performance is determined to have been significantly and favorably impacted by non-interest income It is evident that interest-based activities have become less central to the financial stability and corporate strategy of commercial banks, while fee-based, non-intermediation financial services have gained increasing importance in the examination period The empirical results indicate that an increase in non-interest income is correlated with greater bank performance Although interest from traditional banking operations continues to be the primary source of income for banks, non-interest income from non-traditional banking activities is becoming more significant.

The estimation results of return on assets imply that non-interest income and loans to total assets ratio exert beneficial influences on ROA, whereas bank size, deposit to total assets, cost to total assets, loan loss provision ratio and the annual growth rate of GDP adversely affect the bank’s ROA In terms of return on equity model, loans to total assets and non-interest income ratio are found to positively affect assets, the annual growth rate of GDP and inflation rate have negative associations with ROE Meanwhile, bank size has no statistical meaning in ROE model.

These empirical findings are of expectation given the recent reforms in theVietnamese banking industry, as discussed in chapter four Furthermore, this is a good signal for banks looking to diversify their revenue streams through noninterest-based banking services to become more competitive, reduce risk, and boost profitability.

Recommendations

5.2.1 Recommendations for improving bank performance through non-interest income

5.2.1.1 Policy implications for commercial banks

First, to ameliorate bank performance, banking institutions need to raise awareness about the importance of developing non-interest products and services in enhancing overall business performance and efficiency Banks should build specific strategies to develop non-interest income in each period that are associated with the economic development strategies and orientations of the State Bank and the Vietnamese government In other words, to minimize risks and diversify income sources, organizations should plan to manage the bank's income ratio and avoid relying solely on revenue from credit activities.

Second, commercial banks should establish investment departments with the required business responsibilities as well as develop a risk management process for investment activities in order to create an effective portfolio By doing this, banks can clarify future expectations in order to improve their trading activities and foreign exchange controls.

In terms of customer perspective, banks should create customer-oriented frameworks by offering services that depend on each customer characteristics to offer pay attention to strengthening their links with third parties to provide clients with more products and convenience Accordingly, banks can take advantage of the widespread use of non-cash payment to entice clients to use SMS banking, mobile apps, internet banking, and QR payments Especially, banks can expand their range of customers by utilizing the already-existing payment ecosystems that connect to e-commerce platforms or convenience stores through cooperating with e-wallet operators. Furthermore, banks can increase non-interest income from commissions and fees by promoting cross-selling and up-selling, which may stimulate the demand of customers to sell more products and services that they have not used In cross-selling, for example, banks can suggest customers with savings or current accounts additional banking products like credit cards and internet banking, whereas in up-selling, banks could offer customers the option to purchase an item that is slightly better, like a credit card with more features than the one they are considering.

Next, marketing research campaigns can enable banks to interact with customers and to assess their rivals It is undeniable that customer insights are essential for the success of any business However, since customer needs are undergoing rapid change, banks should design processes from the customer point of view to meet those expectations In other words, banks are critically comprehending the market as well as the consumer desires and requirements by observing their preferences and dissatisfaction For instance, online data collection can be done to collect real time data from consumer blogs, forums, or competitor websites at a negligible cost In fact, banks can extend their brands with the target market by conducting research via internet search engines In addition, digital marketing strategies are also recognized as an effective way for banks to promote products or services These approaches are likely to allow the staff of banks to concentrate on enhancing the customer experience by concerns, and offering customers with specialized experience In this manner, banks can attract new customers while also offering useful resources that improve engagement with existing customers.

Furthermore, Vietnamese commercial banks should make more investments in technology and banking infrastructure to speed up transactions, minimize processing times, and improve client satisfaction To become lean organizations, banking institutions need to create long term technological development strategies in parallel with the development of existing resources Besides, it is apparent that simplified business model plays a crucial part in attracting and retaining clients By raising the level of trading automation, banks could shorten the amount of time consumers spend at transaction offices and branches when not necessary Also, it is essential to strengthen the security of consumer information and accounts As an enormous amount of data is stored on the viral network, an information technology failure can cause unauthorized intrusion and loss of customer accounts, which can halt the bank's operations and damage its reputation.

Finally, increased productivity and efficient resource management enable banks to concentrate their attention on new product and process developments, which can further enhance productivity and efficiency Commercial banks should schedule training and retraining courses for their employees on a regular basis to improve the professional skills and keep them abreast of new technological advancements By harnessing these factors, high-quality human resources can help banks bring products and services to customers more easily, which may boost client satisfaction and the likelihood of customers promoting the bank to their friends, family, and coworkers In addition, banks should develop a clear, reliable, and sustainable brand strategy that can be resistant to alterations in consumer perceptions.

First, the legal strategy and policy systems need to be reinforced to lay the solid foundations and favourable conditions for a flourishing banking industry Particularly, the State Bank of Vietnam should continue to improve the legislative system in order to create an ideal environment for commercial banks to develop their non-interest business activities For example, the authorities should create a sufficient legal corridor for licensing the provision of banking products and services, particularly for new services like currency brokerage, financial consulting, and asset management; at the same time, they should develop in-depth markets supporting financial services such as the money market, interbank market, and corporate bond market; and gradually lay the groundwork for the formation and development of derivative markets.

Second, the operations of securities business, capital contributions, and share acquisitions by commercial banks should be rigorously regulated by regulatory authorities using legal documents Specifically, to give banks a legitimate operating space, policymakers should keep enhancing the legal environment by refining the law on derivatives and foreign exchange markets Regarding securities, capital contributions and share purchases, the legal documentation in Vietnam is currently incomplete and constrained, which makes it possible for banks to violate regulations and conduct high- risk business, leading to ineffective operations Therefore, the management agencies should keep updating these legal agreements to cope with the growing banking business environment.

Furthermore, policymakers should promote an investment-friendly environment for Vietnamese commercial banks There are still several restrictions revealed by state inspection and control activities In fact, the activities of joint-stock commercial banks have recently been exposed to a number of ethical risks that exert detrimental effects on their performance Some measures that can be applied are enhancing inspectors' professional credentials, learning from nations with similar or better financial to the actual situation in Vietnam.

5.2.2 Recommendations for enhancing banking performance through other related factors

In addition to accelerating the credit scale, banks need to strengthen their balance sheets and concentrate on dealing with bad debts To address this issue, banks should improve the administration and classification of debts, monitor their cash flow, periodically check the financial situation of the customers, do research and unscheduled inspections In order to detect and manage credit risks as well as avoid bad debts, banks should also develop standards for grading loans according to the value of the collateral and repayment capacity Additionally, they ought to set rules for allocating provisions for potential losses on loans with high levels of risk Also, selecting clients with sound financial standing is of utmost importance.

Bank managers should determine the optimal equity ratio that the bank needs to raise its equity at the ideal pace Moreover, to attain economic efficiency, bank managers need to improve their management skills to use capital sources rationally, especially by avoiding excessive amounts of unprofitable capital In other words, bank executives should establish a specific capital-raising strategy for each period to raise their capital more efficiently Bank managers should also consider analyzing options for supplementing equity in order to increase both the capital adequacy ratio and bank performance.

Subsequently, banks can specialize in the lending process by analyzing and evaluating credit granting activities for programs and economic sectors By doing research and conducting market analysis in each operating area such as urban,bordering, and rural areas, banks can build a list of credit products that are appropriate for each location and each market segment, thereby selecting distribution channels and organizing the network structure. mobilization, customer deposits are not always inexpensive Hence, banks should have a suitable capital mobilization plan to reduce costs and improve company efficiency, such as by creating scenarios to deal with periods of strong fluctuations in interest rates, enhancing the standard of non-credit services and drawing low-cost demand deposits to boost commercial banks' operational effectiveness in the long term In addition, in order to attract more deposit customers, banks can offer incentives or promotions to customers who deposit large sums of money at the counter, organize gift giving activities for customers on special occasions like holidays and birthdays, encourage customers to use online savings accounts to earn higher interest rates, intensify marketing efforts to advertise goods and services, and widely publicize savings products to customers on various platforms like Facebook, email, websites, and mobile apps.

Finally, to improve operational efficiency, banks need to remain vigilant and keep costs well under control While there are many potential cost management measures, banks need to detect which are the most effective for them For example, to manage staffing costs, banks need to employ tools and processes to manage employee performance and assess training requirements If banks identify any outmoded procedures, they need to be eliminated or replaced with more effective frameworks that focus on the needs of the consumer By cracking down on high-cost areas and implementing reduction strategies, banks can reallocate their resources towards projects and initiatives that provide better earnings than present operations and potentially recapture their declining customer bases.

Limitations and further research directions

Employing annual data from Vietnamese commercial banks over the 2011 -

2021 period, this thesis attempts to do an empirical analysis that investigates the link between non-interest income and bank performance Although the research has achieved specific results as initial primary objectives, it still contains some constraints

First and foremost, due to the restrictive circumstances of information accessibility in Vietnam, many commercial banks either do not publish their financial statements or disclose them insufficiently To put it differently, the chief obstacle that the author encountered was the availability of recent and well-organized data to undertake a thorough estimation of all research variables Thus, data from the entire Vietnam commercial banking system could not be gathered for the study; instead, it was solely based on the information that was found in different published documents. Because of this, the conclusion of this research may be exposed to some limitations in the research scope Second, the indicators that represent the bank performance are not sufficient Specifically, return on assets and return on equity are used in this study, yet there are still many metrics that can be used to assess bank performance, such as net interest margin, return on capital employed or return on investment Furthermore, this study does not examine each specific source of non-interest income, but it just takes the overall amount of non-interest income into account In addition, without considering technical elements, technological advances, the adoption of banks, or market trends, the thesis primarily analyzes the financial aspect of commercial banks to make recommendations for banks Therefore, these variables should be included in further research in order to provide more comprehensive findings.

Following are some directions for future research that can be found in this thesis.Firstly, future studies can increase the number of observations by extending the research period to 2008, which was the year when the global economy experienced a financial crisis Hence, the impact of non-interest income on the performance of commercial banks can be comprehensively evaluated when comparing the bank data before and after the recession period Moreover, in the long-term, more research can be conducted with a more complete set of bank data, comprising institutions that were excluded from this study due to a lack of information It is apparent that the precision of the variables representing the efficiency of operations of banks apart from return on assets and return on equity should also be estimated so as to compare the impact of non- interest income on the performance of commercial banks in different fields. Furthermore, future research articles can separate the non-interest income sources into fee incomes, trading revenues and non-fee incomes to study the impact of noninterest income more specifically Finally, technical elements, technology advances, and the adoption of banks or market trends should also be considered to establish more sufficient research models.

The thesis has made recommendations to improve the bank performance based on the findings of the quantitative models These suggestions are intended to promote the positive effects and mitigate the negative impacts that influence the performance of the Vietnamese banks Limitations and potential directions for future research have also been found in chapter five With the research results, the thesis has provided compelling evidence the direction of the effect of non-interest income on the bank performance.

Hoang Van Cuong, Nguyen Thi Quynh Loan, Tran Manh Dung and Hoang Tuan Dung

2020 The Impact of Income Diversification on Liquidity Creation and Financial Performance of Vietnamese Commercial Banks Accounting, 6 (4), pp, 553-568.

Dang Van Dan 2020 Bank performance and the relationship between lending and fee- based activities in an emerging market Malaysian Journal of Economic Studies, 57(2), 213–231.

Huynh Thi Huong Thao, Pham Thi Kim Anh & Ngo Minh Phuong 2020 Impact of non- interest income on profitability of Vietnamese commercial banks Journal of Banking

Nguyen Minh Sang 2017 Income Diversification and Bank Efficiency in Vietnam,

Journal of Economics and Development 19 (3), pp, 52-67.

Nguyen Minh Sang and Nguyen Thi Thuy Trang 2018 Impacts of non-interest income on risks and profitability in Vietnam’s commercial banks Journal of Banking Science of

Nguyen Minh Sang and Tran Thi Thanh Tam 2020 Analysis of the impact of non- interest income to the operational efficiency of commercial banks in Vietnam.

Nguyen Kim Quoc Trung 2021 Determinants of bank performance in Vietnamese commercial banks: an application of the Camels model Cogent Business & Management, https://doi,org/10,1080/23311975,2021,1979443.

Nguyen Quoc Anh and Tang My Sang 2022 Impact of non-interest income on profitability of Vietnamese commercial banks Journal of Finance and Marketing

Ho Thi Hong Minh and Nguyen Thi Canh 2015 Income diversification and factors affecting profitability of Vietnamese commercial banks Journal of banking technology, no,106 -107, pp,13-23.

Hoang Ngoc Tien and Vo Thi Hien 2010 Discuss the method of calculating non- interest income ratio of commercial banks Journal of banking technology, 48, 36-39.

Nguyen Khanh Ngoc (2019) Revenue Diversification, Risk and Bank Performance of Vietnamese Commercial Banks Risk Financial Management, 12(3),138.

Le Long Hau and Pham Xuan Quynh 2016 Impact of income diversification on business performance of Vietnamese commercial banks Journal of banking technology, no,124, pp,11-22.

Le Long Hau and Pham Xuan Quynh 2017 Impact of non-interest income on the performance of Vietnamese commercial banks in the period 2006-2016 Journal of banking, 9, https://tinyurl,com/29fcbyxm.

Le Hai Trung 2021 Diversification and Bank performance: The case of Vietnamese commercial banks Journal of Economic and Banking Studies, vol 01, pp, 22-34.

Ahmed Raza ul Mustafa, Riaz Hussain Ansar & Muhammad Umair Youni 2012 Does the loan loss provision affect the banking profitability in case of pakistan? Asian

Chen, Y.-K., & Lin, Y.-C 2010 Income and loan perspectives: Do Taiwan banks benefit from diversification? Management Review, 29, 141–145.

Chiorazzo, V.,&Milani, C & Salvini, F 2008 Income diversification and bank performance: Evidence from Italian Banks Journal of Financial Services Research,33(3), 181-203.

Craigwell, R., & Maxwell, C 2005 Non-interest income and financial performance at commercial banks in the Caribbean In Research Department, Central Bank of

DeYoung, R., & Rice, T 2004 Noninterest income and financial performance at US commercial banks Financial Review, 39(1), 101-127.

DeYoung, R., & Roland, K 2001 Product mix and earnings volatility of commercial banks: Evidence from a degree of total leverage model Journal of Financial

Elyasiani, E & Wang, Y 2012 Bank holding company diversification and production efficiency Applied Financial Economics, 22(17), 1409-1428.

Haubrich, J G., & Young, T 2019 Trends in the Noninterest Income of Banks.

Economic Commentary, Research Department of the Federal Reserve Bank of

Hughes, J.P 1999 Incorporating risk into the analysis of production, presidential address to the Atlantic Economic Society Atlantic Economic Journal 27, 1-23.

Hughes, J.P., Lang, W., Mester, L.J & Moon C.-G 2000 Recovering risky technologies using the almost ideal demand system: an application to U.S banking. Journal of Financial Services Research 18, 5-27.

Hunjra, A I., Zureigat, Q., Tayachi, T., & Mehmood, R 2020 Impact of non-interest income and revenue concentration on bank risk in South Asia Banks and Bank Systems. 15(4), 15-25, https://doi,org/10,21511/bbs,15(4),2020,02.

Hsieh, M.-F., Chen, P.-F., Lee, C.-C., & Yang, S.-J 2013 How does diversification impact bank stability? The role of globalization, regulations, and governance environments Asia-Pacific Journal of Financial Studies, 42, 813–844.

Kohler, M 2014 Does non-interest income make banks more risky? Retail-versus Investment-Oriented Banks Review of Financial Economics, 23, 182–193.

Klein, P G & Saidenberg, M R 1998 Diversification, Organization, and Efficiency: i

Evidence From Bank Holding Companies Federal Reserve Bank of New York.

Lee, C., Yang, S., & Chang, C 2014 Non-interest income, profitability and risk in banking industry: A cross-country analysis North American Journal of Economics and

Lepetit, L., Nys, E., Rous, P., & Tarazi, A 2008 Bank income structure and risk: An empirical analysis of European Banks Journal of Banking and Finance, 32, 1452– 1467.

Meslier, C., Tacneng, R & Tarazi, A 2014 Is Bank Income Diversification Beneficial? Evidence from an Emerging Economy Journal of International Financial Markets,

M, Mostak Ahamed 2017 Asset quality, non-interest income, and bank profitability: Evidence from Indian banks Economic modeling, 63, 1-24.

Moudud-Ul-Huq, Changjun Zheng, Anupam Das Gupta & Badar Nadeem Ashraf 2018. Does Bank Diversification Heterogeneously Affect Performance and Risk-taking in ASEAN Emerging Economies? International Business and Finance https://doi,org/10,1016/j,ribaf,2018,04,007.

Nguyen, J 2012 The relationship between net interest margin and non-interest income using a system estimation approach Journal of Banking & Finance, 36(9), 2429–2437. Jacob A, Bikker & Jaap W.B.Bos 2008 Bank performance: A theoretical and empirical framework for the analysis of profitability, competition and efficiency Routledge

International Studies in Money and Banking.

Salike, N., & Ao, B 2018 Determinants of bank’s profitability: Role of poor asset quality in Asia China Finance Review International, 8(2), 216–231.

Sanya, S., & Wolfe, S 2011 Can banks in emerging economies benefit from revenue diversification? Journal of Financial Services Research, 40, 79–101.

Stiroh, K 2004 Diversification in banking: Is non-interest income the answer? Journal of Money, Credit and Banking, 36, 853–882.

Stiroh, K 2006 A portfolio view of banking with interest and noninterest income,

Journal of Money Credit, and Banking, 28, 1351–1361.

Stiroh, K., & Rumble, A 2006 The dark side of diversification: The case of US financial holding companies Journal of Banking and Finance, 30, 2131–2161,

Sherika Antao & Ajit Karnik 2022 Bank Performance and Noninterest Income: Evidence from Countries in the Asian Region Asia-Pacific Financial Markets, 29, 477- 505.

Rayenda Brahmana, Maria Kontesa & Rachel Elfra Gilbert 2018 Income diversification and bank performance: evidence from Malaysian banks Economic Bulletin, vol 38. Rogers Ochenge 2022 The effect of revenue diversification on bank profitability and stability during the COVID -19 Pandemic: Evidence from Kenya KBA Centre for Research on Financial Markets and Policy Working Paper Series, No, 59.

Odesanmi & Wolfe 2007 Revenue diversification and insolvency risk: Evidence from banks in emerging economies Social Science Research Network.

MULTICOLLINEARITY TEST

Pooled OLS regression

Sourc e ss d f MS Numb e r of obs = 285

White’s test

Cameron & Trivedi's decompữsition of IM-test

Fixed effects method

Fixed-effects (within) regression Group varỉable: NH

OA Coefficient std err t p>ltl [95% conf interval]

78733513 (íraction of varỉance due to u_i)

Random effects method

Number Number of obs = of groups =

Wald chi2(10) = 51.84 corr(u_i, X) = 0 (assumed) Prob > chi2 = 0.8000

OA Coeffỉcỉent std err z p>|z| 95% cont ỉnterval]

Hausman test

(b-B) Difference sqrt(diag(V_b-V_B)) std. err.

■ b = Consistent under H0 and Ha; ũbtained írom xtreg

B = Inconsistent under Haj eííicient under H0; ũbtained írom xtreg

Test of H0: Difference in coeííicients not systematic chỉ2(10) = (b-B)'[(V_b-V_B) A (-l)](b-B) = 2.65

Breusch - Pagan LM test

Breusch and Pagan Lagrangian multiplier test for random effects

ROA[NH,t] = Xb + u[NH] + e[NH,t]

Test: Var(u) = 0 chibar2(01) = 678.13Prob > chibar2 = 0.0000

Wooldridge test

Mũũldridge test for autocorrelation in panel data

Collinearity Diagnostics

Feasible Generalized Least Squares regression

Cross-sectional time-series FGLS regressĩon

Panels: generalized least squares heteroskedastic

AR(1) coefficient for all panels (0.6698)

Estimated Cũvariances = 26 Number of ũbs =

= 11 Obs per group: min = 10 avg = 10.96154 max = 11

R ỮA Coefficient std err z p>[z[ [95% conf interval]

Generalized Method of Moments estimation

Dynamĩc panel-data estimation, two-step System GMM

Group variable: NH Number of obs = 156

Time variable : YEAR Number of groups = 26

Number GÍ instruments = 24 obs per group: min

A Coefficien t std. err z p>|z| [95% conf intervaỉ] RO

Warning: Uncorrected two-step Standard errors are unreliable 7

Instruments for first differences equation

D.(L.SIZE L2.NPL L2.CAP L2.LOAN L2.DEP L2.COST L2.LLP L3.GDP L3.INF) GMM-type (missing=0, separate Instruments for each period unless collapsed) L2.(L7.R0A L4.NNII)

L.SIZE L2.NPL L2.CAP L2.LOAN L2.DEP L2.COST L2.LLP L3.GDP L3.INF GMM-type (missing=0, separate Instruments for each period unless collapsed) DL.(L7.R0A L4.NNII)

Arellano-Bond test for AR(1) in first differences: z = -1.00 Pr > z = 0.317 Arellano-Bond test for AR(2) in íirst differences: z = 0.74 Pr > z = 0.457

Sargan test of overid restrictions: chi2(12) = 12.78 Prob > chi2

(Not robust, but not weakened by many Instruments.)

Hansen test of overid restrictions: chi2(12) = 8.63 Prob > chi2

= (Robust, but can be weakened by many instruments.)

Difference-in-Hansen tests of exogeneity of instrument subsets:

Hansen test excluding group: chi2(5) = 5.32 Prob > chi2

6.37 Difference (null H = exogenous): chi2(7) = 3.31 Prob > chi2 8

= 6.85 iv(L.SIZE L2.NPL L2.CAP L2A0AN L2.DEP L2.COST L2.LLP L3.GDP 5

Hansen test excluding group: chi2(3) = 5.17 Prob > chi2

6.16 6 Difference (null H = exogenous): chi2(9) = 3.46 Prob > chi2

Independent Sample T-Test

Two- sample t test with equal variances

Group Obs Mea n std err std dev [95% conf interval]

7 -.0086147 0010175 diff = mean (High NNI) - mean

H0: diff = 0 Degre es of íreedom

Ha: diff < 0 Ha: diff ! = 0 Ha: diff > 0

Pooled OLS regression

Coefficien t std err t p>[t| conf [95% interval]

R-squared - 33 3.42 Adi R-squared - 86 8.43 Root MSE — 77 8648

Cameron & Trivedỉ’s decomposỉtion of IM-test

Fixed-effects (within) regression Number of obs

OE Coefficient std err t p>|t| [95% conf interval]

249) = 5 of variance due to 91 u_i) Prob >

F = 0.0000 landom-effects GLS regression Number of obs = 285 ãroup variable: NH Number of groups = 26 l-squared: Obs per group

R ỮE Cũefficient std err z P>M [95% conf interval]

1990523 (fraction of variance due to u_i)

(b) feE (B) reE (b-B) Difference sqrt(diag(v_b-V_B)) std err.

F 6804865 461352 2191345 0363705 b = Consistent under H0 and Ha; obtained from xtreg

R = Inconsistent under Ha, efficient under H0; obtained from xtreg

Test of H0: Dỉííerence in coefficients not systematic chi2(10) = (b-B)'[(V b-V B)^(-l)](b-B) = 375.83

Modified Wald test for groupwise heteroskedasticity in fixed effect regression model

HO: sigma(i) A 2 = sìgma A 2 for all ì chi2 (26) = 297.72

Appendix 5.7 Wooldridge test nlooldridge test for autocorrelation in panel data

HO: no first-order autocorrelation

Cross-sectional time-series FGLS regression

Correlation: common AR(1) coefficient for aỉl panels (0.6066)

Estimated COVÍ arỉances = Numb er of obs 285

Estimated autocorrelations = Numb er of groups = 26

Estimated coefficients = Obs pe r group: min = 10 avg = 10.9615 max 4

E Coefficient std err z p>[z| [95% conf ỉnterval]

Appendix 5.10 Generalized Method of Moments estimation

Dynamic paneỉ-data estimation, two-step System GMM

Group variable: NH Time variable : YEAR Number of

Number of obs = Number of groups = obs per group: min = avg = max =

E Coefficien t std err z p>|z| [95% conf interv al] RO

Warning: Uncorrected two-step Standard errors are unreỉiable 57

Fixed effects method

Fixed-effects (within) regression Number of obs

OE Coefficient std err t p>|t| [95% conf interval]

249) = 5 of variance due to 91 u_i) Prob >

F = 0.0000 landom-effects GLS regression Number of obs = 285 ãroup variable: NH Number of groups = 26 l-squared: Obs per group

R ỮE Cũefficient std err z P>M [95% conf interval]

1990523 (fraction of variance due to u_i)

(b) feE (B) reE (b-B) Difference sqrt(diag(v_b-V_B)) std err.

F 6804865 461352 2191345 0363705 b = Consistent under H0 and Ha; obtained from xtreg

R = Inconsistent under Ha, efficient under H0; obtained from xtreg

Test of H0: Dỉííerence in coefficients not systematic chi2(10) = (b-B)'[(V b-V B)^(-l)](b-B) = 375.83

Modified Wald test for groupwise heteroskedasticity in fixed effect regression model

HO: sigma(i) A 2 = sìgma A 2 for all ì chi2 (26) = 297.72

Appendix 5.7 Wooldridge test nlooldridge test for autocorrelation in panel data

HO: no first-order autocorrelation

Cross-sectional time-series FGLS regression

Correlation: common AR(1) coefficient for aỉl panels (0.6066)

Estimated COVÍ arỉances = Numb er of obs 285

Estimated autocorrelations = Numb er of groups = 26

Estimated coefficients = Obs pe r group: min = 10 avg = 10.9615 max 4

E Coefficient std err z p>[z| [95% conf ỉnterval]

Appendix 5.10 Generalized Method of Moments estimation

Dynamic paneỉ-data estimation, two-step System GMM

Group variable: NH Time variable : YEAR Number of

Number of obs = Number of groups = obs per group: min = avg = max =

E Coefficien t std err z p>|z| [95% conf interv al] RO

Warning: Uncorrected two-step Standard errors are unreỉiable 57

Hausman test

(b) feE (B) reE (b-B) Difference sqrt(diag(v_b-V_B)) std err.

F 6804865 461352 2191345 0363705 b = Consistent under H0 and Ha; obtained from xtreg

R = Inconsistent under Ha, efficient under H0; obtained from xtreg

Test of H0: Dỉííerence in coefficients not systematic chi2(10) = (b-B)'[(V b-V B)^(-l)](b-B) = 375.83

Modified Wald test for groupwise heteroskedasticity in fixed effect regression model

HO: sigma(i) A 2 = sìgma A 2 for all ì chi2 (26) = 297.72

Appendix 5.7 Wooldridge test nlooldridge test for autocorrelation in panel data

HO: no first-order autocorrelation

Cross-sectional time-series FGLS regression

Correlation: common AR(1) coefficient for aỉl panels (0.6066)

Estimated COVÍ arỉances = Numb er of obs 285

Estimated autocorrelations = Numb er of groups = 26

Estimated coefficients = Obs pe r group: min = 10 avg = 10.9615 max 4

E Coefficient std err z p>[z| [95% conf ỉnterval]

Appendix 5.10 Generalized Method of Moments estimation

Dynamic paneỉ-data estimation, two-step System GMM

Group variable: NH Time variable : YEAR Number of

Number of obs = Number of groups = obs per group: min = avg = max =

E Coefficien t std err z p>|z| [95% conf interv al] RO

Warning: Uncorrected two-step Standard errors are unreỉiable 57

Wooldridge test

nlooldridge test for autocorrelation in panel data

HO: no first-order autocorrelation

Collinearity Diagnostics

Cross-sectional time-series FGLS regression

Correlation: common AR(1) coefficient for aỉl panels (0.6066)

Estimated COVÍ arỉances = Numb er of obs 285

Estimated autocorrelations = Numb er of groups = 26

Estimated coefficients = Obs pe r group: min = 10 avg = 10.9615 max 4

E Coefficient std err z p>[z| [95% conf ỉnterval]

Appendix 5.10 Generalized Method of Moments estimation

Dynamic paneỉ-data estimation, two-step System GMM

Group variable: NH Time variable : YEAR Number of

Number of obs = Number of groups = obs per group: min = avg = max =

E Coefficien t std err z p>|z| [95% conf interv al] RO

Warning: Uncorrected two-step Standard errors are unreỉiable 57

Generalized Method of Moments estimation

Dynamic paneỉ-data estimation, two-step System GMM

Group variable: NH Time variable : YEAR Number of

Number of obs = Number of groups = obs per group: min = avg = max =

E Coefficien t std err z p>|z| [95% conf interv al] RO

Warning: Uncorrected two-step Standard errors are unreỉiable 57

D.(L.SIZE L4.NPL L.CAP L.LOAN L.DEP L.COST L.LLP L.GDP L2.INF)

Standard separate Instruments íor each period unless Cũllapsed) equation

L.SIZE L4.NPL L.CAP L.LOAN L.DEP L.COST L.LLP L.GDP L2.INF

DL.(L7.R0E L.NNII) separate Instruments tor each period unless collapsed)

AR(1) in íirst ditíerences: z = -0.69 Pr > z = 0.491 AR(2) in tirst ditterences: z = -1.35 Pr > z = 0.178

Sargan test ũt ũverid restrictions: chi2(13) = 9.40 Prob > chi2 = 0.742

(Not robustj but not Heakened by many instruments.)

Hansen test ũf overid restrictians: chi2(13) = 5.45 Prab > chi2 = 0.964

(Robust, but can be weakened by many Instruments.)

Difference-in-Hansen tests ot exogeneity ũf instrument subsets: GMM Instruments for levels

Hansen test excluding group: chi2(5)

Ditíerence (null H = exogenous): chi2(8) 3.23 Prob > chi2 = 9 0.91 iv(L.SIZE L4.NPL L.CAP L.LOAN L.DEP L.CŨST L.LLP L.G 9

L2.INF Hansen test excluding grũup: chi2(4) )

Ditterence (null H = exogenous): chi2(9) 4.73 Prob > chi2 = 0 0.85