Corporate social responsibility expenditure and financial performance: A comparison of Vietnamese listed and unlisted banks

Abstract There are legal requirements for a bank to be officially listed on the Vietnam stock exchange; hence by the end of 2020, Vietnam’s banking industry had 43% of banks officially listed on the stock exchange and 57% of banks not officially listed. Accordingly, the difference between these two groups of banks is clear. This study aims to find empirical evidence for a one-way relationship between corporate social responsibility expenditure (CSRE) and the financial performance of Vietnamese commercial banks. In addition, the study also looks for the influence of each component of CSRE, including Community, Government, and Employee Responsibility, on the financial performance of listed and unlisted banks. Using the Generalized Method of Moments (GMM) controlling for heteroscedasticity, serial correlation, and endogeneity effects, the research results of the study show that CSRE as a cost reduces the bank’s financial efficiency; however, social responsibility spending with the community is a positive factor that increases the bank’s financial efficiency through creating an image and good impression with society and customers. Especially significantly stronger for commercial banks that have been officially listed on the stock exchange. Research results show the need to strengthen CSR implementation in the banking industry.


Introduction
In recent years, researchers have long been interested in looking at the precursors of what has come to be called Corporate Social Responsibility (CSR). The CSR journey began centuries ago and is now evolving at an unprecedented rate in both conceptual and empirical research with no sign of slowing down (Quazi et al., 2016). The concept of corporate social responsibility has emerged as one of the essential concerns for managers and has become the subject of intense academic debate (Ehsan et al., 2018). Implementing CSR is one of the actions that are important to the success of an organization because customers like the organization to be economically and socially responsible. Corporate social responsibility disclosure improve financial performance in that responsibility. To fully meet corporate social responsibility, businesses need a process to integrate social, environmental, ethical, human rights, and consumer concerns into their business operations and strategy core with close cooperation with stakeholders (EC, 2011).
Another definition that is more common and often used to help organizations develop CSR policies is the concept of the United Nations Global Compact (UNGC), introduced in 2010. Accordingly, CSR is a strategic policy initiative for businesses committed to aligning their operations and strategies with ten principles in human rights, labor, environment, and anti-corruption (UNGC, 2010). Since the UNGC issued the ten CSR principles, nearly 12,000 businesses worldwide have committed to adopting sustainable and socially responsible policies around human resources rights, labor, environment, and anti-corruption (Moshkin, 2019).
In 2010, the International Organization for Standardization (ISO) released voluntary guidance ISO 26,000 on social responsibility. The content of social responsibility includes seven core factors: (i) Corporate governance to ensure benefits for shareholders; (ii) Good implementation of human rights; (iii) Good relations and treatment with employees; (iv) Environmental protection; (v) Fairness in operations; (vi) Ensuring the interests and safety of consumers; (vii) Contributing to the social community. This is a popular guide to help organizations carry out CSR activities (Alpana & Alpana, 2014). At the same time, scholars also use it to design CSR measurement topics; the research of T. H. Y. Tran (2016) in Vietnam was an example.
A bank is also a type of business with a unique feature of providing financial products: Customers will decide to "buy" (deposit and borrow) based on the "price" of the product, which is "interest" (T. H. Y. Tran, 2016). Furthermore, the view of CSR in the banking industry is not different from that of enterprises. Regarding environmental issues, banks are often considered low-emissions and environmentally friendly businesses (T. T. Tran et al., 2016). The bank does not create harmful chemicals or release pollutants into the air, soil, or water (Thompson & Cowton, 2004). However, banks can also pollute indirectly by lending money to polluting companies or projects, indirectly causing significant environmental damage (Zhang et al., 2011). Based on Carroll theory and ISO (2010) concept and Stakeholder theory, a bank's social responsibility is understood as the responsibility of commercial banks for the activities and impacts of commercial banks' decisions on the environment, employees, and the community through transparency and ethical behavior in order to contribute to sustainable development, taking into account the wishes of stakeholders, following the law, consistent with international standards of conduct and integrated into all activities and practices in relationships of the bank.

Theories explaining corporate behavior related to social responsibility
A wide range of theories to explain corporate behavior related to the performance of social responsibility, such as Slack resource theory, Stakeholder theory, Legitimacy theory, Shareholder theory, and Carroll theory.
Slack resource theory has been the essential foundation for research on CSR expenditure. Slack resource theory emphasizes that when firms have abundant financial resources (such as when firms are highly profitable), firms are financially sound and therefore willing to pay more to society. The firm's ability and propensity to engage in social involvement rises along with the increase of their available financial resources (Seifert et al., 2004). Some see CSR as an agency cost because socially responsible spending is misdirected by managers who prefer diversifying corporate earnings away from their rightful owners to social (Friedman, 1970;McWilliams et al., 2006); thereby enhancing the reputation of the firm's managers with society.
Stakeholder theory is considered one of the central theories that lay the foundation for developing research on social responsibility. A company is addressing the needs of its shareholders (owners) and other constituencies, who form a powerful force for driving the company's operations. When stakeholders are satisfied, it helps firms create competitive advantages and improve financial performance (Freeman & Evan, 1990).
Carroll Theory shows which activities of enterprises are considered CSR. According to Carroll (2016), CSR is divided into four components such as economic, legal, ethical, and philanthropic. This set of four responsibilities creates a foundation to help define some details and shape (describe) the nature of corporate responsibility to the society in which it participates. Based on the pressure of society expressed through the degree to which businesses are expected to fulfill these responsibilities. Carroll has arranged the above four responsibilities from bottom to top like a pyramid. The bottom is economic responsibility, considered the most fundamental and expected, followed by legal and moral responsibility. At the top, charitable responsibility is the most diminutive expected meaning.

Measuring corporate social responsibility
There are many methods of collecting and measuring CSR scores, of which four are standard methods used by many studies Rating index dataset, Content analysis, Survey data, and Monetary data.

Rating index dataset
One-way or multi-dimensional rating index datasets are available databases made and hosted by independent rating agencies such as Asset 4, Bloomberg, EIRIS, KLD, Sustainalytics. . . The main advantage of datasets is availability, thus minimizing collection effort. In addition, companies are comparable as they are evaluated by the same criteria for the entire database (Galant & Cadez, 2017). The main disadvantage of datasets is that they are not an ideal measure of CSR as private companies with agendas often compile them. They may not necessarily use the rigorous methods normally expected in scientific research.

Content analysis
Content analysis is a versatile method that is performed by measuring the disclosure of CSR activities in company articles, most often in an annual report or a separate CSR disclosure. Content analysis typically includes: (i) Identifying structures of interest-the corporate CSR components; (ii) Finding information about these structures; (iii) Systematizing qualitative information to generate quantitative scores by encoding the information. The main advantage of this method is its flexibility for the researcher. A researcher can specify the CSR components of interest, collect data according to those components, and encode the data numerically for research purposes (Galant & Cadez, 2017). The second advantage is that it can be easily implemented in studies with large sample sizes as CSR information has become more accessible due to the increasing attention paid by companies to the disclosure of responsible practices and social responsibility (Leaniz & Bosque, 2013). That is the reason why this method is widely used in CSR literature. However, the content analysis method also has its weaknesses. The main weakness of this approach is that the researcher's subjectivity is embedded in all stages of the research process, from selecting CSR aspects of interest in data collection, data interpretation, and data encryption (Galant & Cadez, 2017). Second, content analytics captures only what the company has done, which may differ from its current performance. Third, there is still no consensus on the scales to measure CSR activities. In addition, this method does not capture the quality of CSR disclosure but focuses more on the quantity of disclosure.

Survey data
A questionnaire-based survey is often used when a rating agency does not rate a particular company, and the company's reports are not available or sufficient for meaningful analysis. In such cases, researchers need to collect baseline CSR data by sending questionnaires to knowledgeable respondents or interviewing them (Galant & Cadez, 2017). The researchers then analyzed the responses received and assessed the level of CSR. According to Galant and Cadez (2017), the main advantage of the survey method is similar to that of content analysis. It provides excellent flexibility for the researcher to identify CSR components of interest and collect data on them.
In contrast, the main problem with using survey tools in CSR studies is that it is challenging to form well-structured questionnaires to obtain the required CSR data. Second, the answers received only represent the opinions and perceptions of the respondents so that they may be judgmental, internal to the company, and not the perception of the entire company or parties other related. As such, the answers may be biased (Gbadamosi, 2016). In addition, this method has some costly and time-wasting disadvantages.

Monetary data
CSR can be measured by spending levels such as company spending on Community Responsibility (Public Welfare Undertaking, philanthropic giving, Community Involvement. . .), Employee Responsibility (Fair Employment, Health & Safety, Training and Education, Professional Development. . .), and government responsibility (tax, government charges) (Iqbal et al., 2014;Madugba & Okafor, 2016). Monetary data are often used because they are readily available, easily obtainable, and inexpensive. They are obtained directly from the financial statements or annual reports without needing collection, processing, and comparison (Carroll, 2016). Using economic data to represent CSR, although not fully reflective of the business's process of implementing and disclosing information, does reflect quite well the part of expenditures spent on charity, social work, and programs, study promotion programs, costs of environmental protection, public health, contribution to the State budget, improvement of employees' income, and influence on the financial performance of enterprises. When using financial data to measure CSR, studies often identify with the concept of corporate social responsibility spending or Corporate social responsibility expenditure.

Corporate social responsibility expenditure and financial performance
Conclusions about the impact of social responsibility spending on banks' financial performance in published works vary widely, with some studies finding positive, adverse, mixed effects, or no significant relationship. For instance, Adewale and Rahmon (2014) examined the impact of CSR spending on the financial performance of banks in Nigeria over 20 years, from 1990 to 2010. Their research results show a positive relationship between CSR expenditure and earnings after tax. The study of Kabir and Qayum (2016) used return on equity, investment, deposit, ROI, ROA, Net profit, EPS, and P/E ratios as bank's financial performance variables to examine the relationship between CSR expenditure and Islamic banks' financial performance in Bangladesh. Their results show a significant positive impact of CSR expenditure on Islamic banks' deposits and investments. The other six variables: ROE, ROA, ROI, net profit EPS, and P/E ratio, are not significantly correlated with CSR expenditure. Shukla (2017) uses Profit After Tax (PAT), Return on Assets (ROA), Return on Equity (ROE), and market capitalization to proxy of bank's financial performance. Regression analysis results show a positive relationship between CSR expenditure and different measures of financial performance while the firm size is controlled. Oladele and Mokuolu (2020) apply panel data analysis as a primary estimation technique. The results showed that although CSR expenditure of enterprises had a positive effect on their performance, the effect was not significant. Research has also shown that the banking sector is more organized and unique in its approach to CSR and its implications for its performance than oil and gas companies. It is recommended that companies in Nigeria find ways to make their CSR spending positively and significantly impact their performance. Malik et al. (2019) investigated the role of peer pressure on banks' CSR activities and the long-term impacts of their CSR spending on financial performance. They found that the bank's CSR spending increases with that of its peer banks, and there was no association between a bank's CSR spending and the spending of its non-peer banks. The additional study suggests that a bank's CSR spending increases not only its current profitability but also its future profitability. Bani-Khaled et al. (2021) examined the relationship between CSR expenditures and the financial and non-financial performance of Jordanian commercial banks during 2008-2018. The study found a positive, statistically significant relationship between CSR expenditures and return on equity (ROE). For banks' non-financial performance, CSR expenditures positively impact total deposits and training expenditures.Studies showed that CSR expenditure negatively impacts bank financial performance less than favorable. However, many recent studies have found a negative effect of CSR expenditure on company's profitability as Su et al. (2020), Sharma and Aggarwal (2022), Lopatta et al. (2022). Implied results the mandatory CSR expenditure acts as a burden onto the ongoing activities of the firms and in this case the component of the CSR expenditure must be analyzed to clarify the source of the results. Many studies show that, the impact of human resource expenditure on financial performance is negative implying businesses need to be careful when spending in human resources. However, CSR investments in the community are positively related to financial performance firms, hence the firms can focus the CSR efforts on community issues like campaigns and volunteer programs, donating to charity can help firms build harmonious relationship with the community and benefit (Su et al., 2020). Research results by Raihan et al. (2015) show that spending on disaster recovery, Education, Health, Sports, Arts and Culture, Environment, and other similar activities are negatively related to ROE but positively related to deposits. Research suggests that bank administrators should spend more on CSR to improve customer deposits positively. Madugba and Okafor (2016) carried out a study with a sample of Nigeria's listed banks in the period 2010-2014, using the bank's donations and charity to measure CSR spending; the regression results show CSR spending has a negative relationship with EPS and customer deposits, while there is a positive relationship with ROE. Research by Bani-Khaled et al.
(2021) on a sample of 13 Jordanian commercial banks in the period 2008-2018, in addition to the positive impact of CSRE on ROE, the study also found a negative relationship between CSRE with return on assets (ROA) and Tobin's Q model.

Employee responsibility and financial performance
Corporate social responsibility has become an essential and indispensable strategy in the strategic management of all businesses, no matter what industry they operate in. A business can only survive if it can meet its stakeholders' needs, which can significantly affect the company's bottom line. Among the main stakeholders, employees are an internal factor, so the corporate social responsibility towards employees is the core content, an essential factor constituting the sustainable development of the enterprise. CSR to employees plays a significant role because it attracts suitable and highly qualified labor resources, leading to increased productivity and reduced production costs for businesses (T. H. Tran, 2019), and the possibility that social responsibility spending for employees affects the financial performance of enterprises.
In banking activities, unlike other business activities, a bank's products, and services, are associated with the distribution and use of capital, financial advice, and activities that depend heavily on the trust and confidence of customers. Also, because the production process is associated with the consumption process, the staff plays a significant role in "selling," deciding the success or failure of the bank. Therefore, the responsibility of commercial banks toward employees is even more critical. Several previous studies have found empirical evidence concerning the impact of spending on CSR employees on bank performance. Gbadamosi (2016) Iqbal et al. (2014) (2016) show the positive impact of community responsibility on financial performance. Some studies show a negative correlation between corporate responsibility and financial performance. Research by Gbadamosi (2016) suggests community responsibility is negatively correlated with accounting profit. The study of Buallay (2019) study in the Europe banking sector concludes that community responsibility reduces financial performance as a proxy by ROA, ROE, and Tobin'Q. The study of P. H. Le (2020) shows that Vietnamese commercial banks' community responsibility has a negative impact on financial performance. (2019) use linear and non-linear approaches to examine the impact of community responsibility on the NIM of banks in Poland. The results of the linear models show that there is no significant relationship between community responsibility and NIM. However, further analysis with non-linear models revealed an inverted U-shaped relationship between Community responsibility and NIM.

Government responsibility and financial performance
According to Stakeholders theory, the State is also an essential stakeholder in the operation of enterprises. Uddin et al. (2008) believe that payable taxes are not expenses but a part of the enterprise's contribution to society or social responsibility to the State. Tax avoidance or evasion by businesses is harmful to society and means that businesses do not want to share their success with society. KsiężaK and FischBach (2017) argue that corporate tax is a fair societal adjustment. Because the State requires tax on profits, profitable businesses must pay corporate income tax, just as individuals with income must pay personal income tax. The State spends the tax revenue on the people through social welfare.
Many previous studies have used tax obligation as a representative of CSR to the State. For instance, Khurshid et al. (2016)  shows that community responsibility has no significant impact on the profitability of enterprises that ROE, ROA, and EPS represent. In addition, the study also analyzes the impact of CSR and its components on financial performance by industry. Regression results for Mining, Metal Fabrication, Gas and Water Related Industries, and Oil Production Equipment concluded that community responsibility did not affect the company's financial performance. In contrast, for the Oil and Gas industry, community responsibility affects EPS, with the implication that taxes and philanthropy positively impact profits. One possible explanation for the result is that most oil and gas companies are state-owned, reaping huge profits and attracting vast public attention. Community responsibility can improve consumer loyalty and coordination with local communities during mining.

Research model
This study also uses ROA to measure a bank's financial performance because it is one of the primary metrics of bank profitability (Esteban-Sanchez et al., 2017; Wu & Shen, 2013; Wang et al., 2013). For the independent variable representing the CSR expenditure, the study measure sum of spending on three items: employees, community, and government. Employees' responsibilities include salaries and wages, spending of employees, donations, health, education, training expenditures, and employees' welfare fund (Kakakhel et al., 2015;Kiran, 2015). Community responsibility includes all charitable giving, contributions to society, public welfare undertaking, and community involvement (Esteban-Sanchez et al., 2017;Raihan et al., 2015); The taxes that banks pay to the state are considered a government responsibility (Khurshid et al., 2016;Zhou et al., 2021). Accordingly, we use the following formulas, respectively: The research model of components of CSRE includes community responsibility, employee responsibility, and government responsibility, affecting Vietnamese commercial banks' financial performance.

Estimation method
The research population for this research includes 28 banks Vietnamese commercial banks (excluding Joint venture Banks and Banks with 100% foreign capital). Annual reports and financial statements of banks used for analysis. As a result, there were 28 fully publicized banks from 2011-2019. The variables in this study were collected from 2011 to 2019 and are continuous variables. We do not use data in 2020 because it is affected by the COVID-19 pandemic. Vietnamese banks temporarily suspend operations due to COVID-19 prevention requirements. Therefore, the final study sample is 28 banks, of which 12 are officially listed on the stock exchange, 16 are not officially listed, and the number of observations is 218. First, the study examines the unit root of each variable. It explores the panel's data characteristics, ensuring that each variable used in the study is stationarity before proceeding to the panel co-integration test. Next, the study examines the model's defects, such as multicollinearity, autocorrelation, heteroscedasticity, and endogenous variables. In many studies on factors affecting financial performance (ROA), it has been shown that bank-specific variables such as Non-performing Loan (NPL), bank size (SIZE), and capital are endogenous variables (Athanasoglou et al., 2008;Siddique et al., 2021). Furthermore, when any study GMM model applies, it can remove the problem of endogeneity by introducing some instrumental variables (Siddique et al., 2021 The GMM method has two alternative estimators, Dif-GMM and Sys-GMM. This study chose to use Sys-GMM because it has been improved based on the Dif-GMM version to give a better estimate. The two-step estimator was also chosen because it is more efficient than the single-step version, especially for the Sys-GMM estimate (Huynh et al., 2021).
Before discussing the estimation results, the study was conducted to test the suitability of the regression using the Sys-GMM method. F test to check the statistical significance of the estimated coefficients. If P.value<1%, the estimated coefficients are statistically significant. AR test to determine whether there is a correlation in the model residuals. The model has no quadratic autocorrelation if the AR (2) test has a P.value>10%. Sargan test to check for excessive constraints  (Ngo et al., 2020). Finally, when the number of instruments is less than or equal to the number of groups, it is concluded that the instrumental variables are not weak. Table 2 describes statistics based on a sample of 28 commercial banks; the study period is from 2011 to 2019. The maximum value of ROA is 2.90%, the minimum value is zero, and the mean value is 0.78%. There is a significant difference in financial performance among banks in the sample. The highest value belongs to Techcombank, a listed bank, and the lowest value belongs to the unlisted banks.

Descriptive statistics of research samples
The average CSRE of Vietnamese commercial banks is 14.69%. The lowest score is 2.19%, and the highest is 118.47%. This shows that all banks in Vietnam have CSR expenditure; the minimum value of CSRE is not zero. However, the minimum value of zero of the COM variable and GOV variable shows that some banks have not contributed a single penny to society. Many banks are doing business at a loss, so income tax is 0 VND, while Baovietbank's highest tax payment in 2014 was 35.66%.

Unit root test
All variables in Equation (2) must be stationary before panel data can be analyzed; the study uses the ADF fisher test, which will remove cross-sectional means by using demean. Table 3 presents the results of the panel unit root test of each variable in the model. Table 3 shows that all four tests strongly reject the null hypothesis that all the panels contain unit roots. The simulation results of Choi (2001) suggest that the inverse normal Z statistic offers the best trade-off between size and power and recommends using it in applications. The study has observed that the inverse logit L* test typically agrees with the Z test. Under the null hypothesis, Z has a standard normal distribution, and L* has a t distribution with 5N + 4 (144) degrees of freedom. Low values of Z and L* cast doubt on the null hypothesis (xtunitroot test-Stata.com).   Table 4 presents the correlation matrix of the variables, revealing that these variables are not highly correlated. In addition, the variance inflation factor (VIF) values were below 4, implying that the multicollinearity conclusion was small and did not affect the regression results. In addition, the correlation matrix also shows that the pairwise correlation between the CSRE variable and the ROA variable is negative; The correlation between the GOV variable has a positive correlation with ROA, and the EMP variable has a negative correlation with ROA. No correlation was found between the COM variable and the ROA variable.

Estimation results
6.4.1. Impact of total CSR expenditure on financial efficiency  (2) test has a P.value of more than 10%, which means that the model has no second-order correlation. The Sargan test results show that the model is correct and the variables are reasonably represented. The P.values of the Hansen test of all two models is greater than 10%, indicating that the variables selected as instrumental variables are reasonable. Finally, the number of instruments in the two models is less than or equal to the number of groups, so the instrumental variables are not weak.
Total CSR expenditure, variable CSRE, has an estimated coefficient of −0.0176 and is significant at 1%, concluding that CSR expenditure tends to reduce the FP of banks. This result implies that the total spending of the bank for the community, income tax paid to the State, and spending for employees negatively affect the bank's profit. Banks are spending more and more on CSR but are low efficient. Hence, shareholders and other external stakeholders consider identifying over-or underinvestment in CSR relating to agency issues (Lopatta et al., 2022). The research results are supported by the study of Su et al. (2020), Sharma and Aggarwal (2022), Lopatta et al. (2022). Then, we followed the direction of previous studies to evaluate the impact of each component of CSR expenditure on FP. When estimating the component of CSR expenditure (model 2), it is shown that employee responsibility expressed in wages paid to employees (EMP) is a cost in the bank's business activities so that when wages for employees increase, bank profits will decrease. Community responsibility (COM) has a positive effect on bank performance. This result encourages banks to increase charity activities, environmental protection expenses, scholarship sponsorship, and social activities because these are good deeds that can be useful to a bank's financial performance by improving the bank's image or reputation. Government responsibility through the amount of income tax payable to the State (GOV), the study did not find statistical significance in this variable. The research results in model 2 explain the influence of the total CSR expenditure on FP in model 1: the strongly negative influence of CSR spending on employees is greater than the positive effect of CSR spending on the community. This led to the total CSR expenditure negatively affecting Vietnamese banks' financial performance. Another meaning from the research results in model 1 and model 2 is that the bank's CSR expenditure is mainly the cost of staff salaries.
In contrast, the expenditure on the community is relatively small. However, banks need to see that spending on the community is a positive factor for the banking business because it will increase financial efficiency. Therefore, to improve financial efficiency, one of the solutions is that banks should increase spending on social activities.

Difference between the listed bank and the unlisted bank
According to regulations for Vietnamese joint stock credit institutions to be approved by the State Bank to list on the stock market: (1) having operated for at least 02 (two) years by the time of the request; (2) the actual value of charter capital up to the request time is not lower than the legal capital according to current regulations; (3) profitable business activities based on audited consolidated financial statements and audited separate financial statements for 02 (two) consecutive years preceding the year of *, **, *** represents 10%, 5%, and 1% significance level. application; (4) comply with the restrictions to ensure safety in the operation of the credit institution as prescribed and the instructions of the State Bank for these regulations continuously for 06 (six) consecutive months before the time of proposal; (5) the bad debt ratio is less than 3% of the total outstanding debt at the end of the quarter during the 02 (two) consecutive quarters before the proposed quarter; (6) classify debts and make provision for risks according to regulations of the State Bank at the end of the quarter immediately preceding the requesting quarter; (7) during the 12 (twelve) months preceding the application, the joint-stock credit institution shall not be sanctioned for administrative violations in the monetary and banking sectors in the form of fines with an acceptable amount from 30 (thirty) million VND or more; (8) the time of the request, the Board of Directors, the Control Board of the joint-stock credit institution must have the number and structure to ensure the provisions of current law; (9) at the time of the request, the joint-stock credit institution has an internal audit department and an internal control system to ensure compliance with Articles 40 and 41 of the Law on Credit Institutions and relevant regulations of current law.
Unlisted banks will be those that do not meet some or all of the nine regulatory requirements, one of which is the requirement of financial performance. Financial performance and CSR expenditure statistics between two groups of listed banks and unlisted banks on the Vietnamese stock market in the three years 2017 -2019 (Table 6) show that listed banks consistently have a larger ROA than unlisted banks. The total CSR expenditures of the two groups are similar, but the components of CSR expenditures are different. Listed banks spend more community responsibility and pay more taxes than unlisted banks. Meanwhile, unlisted banks spend on more employees, which shows that unlisted banks are less efficient in using labor than listed banks. *, **, *** represents 10%, 5%, and 1% significance level.
To examine how the impact of CSR expenditure on financial performance is different in listed banks compared to unlisted banks, the study uses the interaction variable between CSRE variable, COM variable, GOV variable, EMP variable with dummy variable LB respectively. The dummy variable LB takes the value one if the bank is officially listed on the Vietnamese' stock market and takes the value 0 otherwise. The estimated results are presented in Table 7.
In Table 7, Models 3 show the interactions between CSRE and LB, whereas Models 4, 5, and 6 include all interactions between COM, GOV, EMP, and LB.
As Model 3 shows, the effects of CSRE still have a negative and statistically significant impact on ROA, as shown in model 1 in Table 5. However, accounting for the interaction between CSRE and LB, a different perspective: while the direct impact of CSRE on FP is still negative, the interaction of CSRE and LB has a positive and significant impact on it. This result indicates that when a bank list officially on the Vietnamese stock exchange, the negative relationship between CSRE and FP is weaker. This result implies that the CSR spending of listed banks is more valuable than those of unlisted banks.
Based on Model 5, the impact of COM on FP is still positive, while in Model 4, the coefficient for the interaction between COM and LB is negative and significant. This negative interaction suggests that the positive relationship between spending toward community and FP becomes more substantial with listed banks. As Khurshid et al. (2016) argued, sacrificing social activities for profit maximization gradually becomes an unhealthy direction for enterprises' long-term survival. Shortterm growth by increasing selling prices is no longer strategically rational when businesses have to deal with pressures from stakeholders and society to be socially responsible as good citizens. The article's findings provide evidence for bank managers that investing in CSR for the community not only creates a good image but also brings financial benefits. Therefore, managers should include CSR as part of strategic plans for growth and profit maximization and serve the personal interests of all stakeholders and not affect collective interests.
In Model 2 Table 5, the coefficient for GOV is insignificant; however, in Model 4 and 6 Table 7, fortunately, GOV is positively related to FP and significant. The coefficients of the interaction variable GOV and LB is positive (Model 5 Table 7), indicating that listed banks have more responsibility to the State than unlisted banks. The amount of tax paid to the State budget also shows that the financial efficiency of listed banks is better.
The EMP is still negative and statistically significant impact on ROA in all models. Spending on a bank's employees is one of the operating expenses; thus, it reduces a bank's financial performance. The estimated coefficients of EMP.LB is positive and significant (Model 6 Table 7). This negative coefficient for the interactions indicates that the negative impact of employee spending on financial performance is weaker when a bank is officially listed.  LB= listed status of bank, take 1 if listed bank, 0 for otherwise. *, **, *** represents 10%, 5%, and 1% significance level.

Conclusion
This study calculates total CSR expenditure, including spending on Community Responsibility (Public Welfare Undertaking, philanthropic giving, Community Involvement, etc), Employee Responsibility (Fair Employment, Health & Safety, Training and Education, Professional Development, etc), and government responsibility (tax, government charges). Then, we consider the relationship between total CSR expenditure, spending toward community, government, employees, and financial performance that is moderated by the listing status of Vietnamese commercial banks. The article found meaningful results by using the annual reports of 28 Vietnamese commercial banks from 2011-2019. First, the impact of total CSR expenditure on the bank's financial performance is negative, and the negative impact of CSRE on FP weakens when the bank is officially listed on the stock exchange. Second, CSR spending for the community positively affects financial performance, and this positive effect in listed banks is more robust than in unlisted banks. This finding is an incentive for banks to increase spending for the community because it helps the bank to have a good image of customers and society, thereby increasing financial efficiency. Third, the impact of government responsibility on financial performance is positive, and the listed banks have more responsibility to the State than unlisted banks. Fourth, CSR spending for the employee harms financial performance, and this effect is weaker with listed banks. Our results provide important policy implications: Vietnamese commercial banks can improve their banks' financial performance by increasing social responsibility spending with the community. This issue is more critical for officially listed banks. Money spent on community increase ROA because this application improves the image and reputation, attracts investors, reduces the cost of capital, and improves relationships with stakeholders to gain favor and approval. For spending on human resources, Vietnamese banks, especially unlisted banks, need to examine the income policies to retain stable employees, motivate them to work effectively, and attract qualified employees. Unfortunately, CSR disclosure requirements are not strongly required by current Vietnamese government regulations, so the expenditures to society of banks are not detailed in their financial statements. It is presented optionally in the notes to the financial statements. Therefore, data collection on COM variables has difficulties and gets few observations in the research model. This problem is a limitation of the study; it suggests that there should be a way to measure CSRE by Vietnam's regulations on CSR disclosure or when regulations on CSR disclosure are clear and mandatory