CEO overpower and corporate social responsibility of commercial banks: The moderating role of state ownership

Abstract This study examines the impacts of powerful CEOs and state ownership on commercial banks’ corporate social responsibility (CSR) in Vietnam, a transitional market in Asia. Given the differences between emerging and developed markets in terms of their institutions and governance, it is essential to explore the impact of CEO power on CSR disclosure in emerging markets. This study collects data from 37 Vietnamese commercial banks from 2010 to 2020. We employ a dynamic system Generalized Method of Moments to overcome endogeneity and heterogeneity issues. The findings show that powerful CEOs negatively reduce CSR programs. CEO power tends to focus less on CSR investments because CSR expenses reduce the operating free cash flow. Meanwhile, our findings indicate a positive relationship between state ownership and CSR developments. This study reports a moderating role of state ownership in empowering powerful CEOs to develop CSR programs in commercial banks. We also perform a robustness test to confirm the persistence of our main findings across subsamples by CEO ages. Our robustness test results indicate that CEOs have the lowest motivation to improve CSR when their ages are from 40 to 60. Our findings align with agency theory, stakeholder theory and prior literature. Finally, our study contributes practical implications for management and policymakers to develop CSR programs sustainably.


Introduction
Corporate Social Responsibility (CSR) has recently become the most significant and contentious corporate governance trend. CSR is an extension of firms' efforts to foster effective corporate governance, ensuring sustainability via sound business practices that promote accountability and transparency. Ullah et al. (2019) argue that since the choices and motives influence CSR disclosure and values of those involved in formulating and making decisions in the organizations, consideration of corporate governance mechanisms, particularly ownership structure and board composition, could be important determinants. Jo and Harjoto (2011) report that efficient CSR programs depend on corporate governance mechanisms, such as board independence, board leadership, and institutional ownership.
Over the past decade, several studies have examined the linkage between integrated reporting system and CSR disclosure. Fifka (2013) provides a comprehensive review of empirical literature regarding CSR reporting and its determinants. Frias-Aceituno et al. (2014) studied the explanatory factors of integrated sustainability and financial reporting. In addition, Frias-Aceituno et al. (2013) analyze the impact of a country's legal system integrated reporting system. Furthermore, Pineiro-Chousa et al. (2019) investigate the effect of financial development on the adoption of standards for reporting practices among developed, emerging, and frontier markets. However, most social responsibility issues are investigated in developed countries. The practice of CSR in developed and emerging countries is different, so there is still a gap in the research on CSR in frontier and emerging economies. Furthermore, Pineiro-Chousa et al. (2019) state that as differences between emerging and developed markets in terms of their institutions and governance, it is essential to explore the impact of CEO power on CSR disclosure in emerging markets. Therefore, this study needs to explore CSR in the context of a frontier economy, specifically the banking industry in Vietnam.
Vietnam is a transitional country with economic growth relying on fossil fuels, so facing environmental problems is becoming more significant. Therefore, the Vietnamese government introduces effective environmental legislation and management policies to promote CSR implementation and corporate disclosure with environmental responsibility. Our sample statistics report that the average CSR score of commercial banks in Vietnam is 3.12, much higher than in neighboring countries. The banking industry actively participates in green credits, which finance environmentally friendly projects. However, banks can indirectly pollute the environment by providing loans for ecologically unfriendly projects. Therefore, developing CSR in the Vietnamese banking industry is necessary for sustainable growth. structure relationships. D. Li et al. (2016) argue that state ownership is vital in fulfilling national macro-control policies, guaranteeing market supply, and implementing major national development strategies. Governments encourage firms to help them achieve social and political objectives. Therefore, higher state ownership implies positive incentives for CSR programs (D. Li et al., 2016).
Although Vietnam has been transitioning from a centralized economy to a market economy, the role of the government remains apparent via state ownership in listed firms. There are many enterprises with 100% charter capital held by the State and joint-stock companies with more than 50% share capital of the State, and the banking industry is no exception. The big four state-owned commercial banks have a significant role in supporting the government in implementing monetary policies. These banks are motivated to increase social responsibility and disclosure to meet government objectives (Lau et al., 2016). Moreover, our sample statistics report that the average state ownership in the Vietnamese banking sector is 9.6%, implying an essential role of the government in the banking sector. Therefore, we hypothesize that higher state ownership develops the CSR of commercial banks in Vietnam.
In addition, it is essential to discuss the moderating role of state ownership in empowering the impacts of powerful CEO on CSR. It is because CEOs have the main power to control the enterprise's operations and to provide essential policy guidelines to understand CSR better. Sheikh et al. (2018) have emphasized that state ownership is positively associated with CEO compensation. A much more significant government influence over business operations will significantly boost the management efforts to meet government expectations. Government shareholders have reason to redirect capital to achieve social goals, contributing to improving CSR. High levels of state ownership encourage CEOs to achieve social goals such as infrastructure development, financial hardship, and unemployment. As a result, these social or political objectives hinder companies from implementing CSR. Therefore, we conjecture that state ownership has a moderating role in empowering powerful CEOs to develop CSR programs in commercial banks.
There needs to be a consensus on the definition of CEO power in the literature. Prior studies have used measures of CEO power that only reflect one or two sources of power, such as the share of the CEO's salary and the duality of the CEO (Jiraporn & Chintrakarn, 2013;D. Li et al., 2016). However, Finkelstein (1992) argues that power is a multidimensional concept and should be measured as an index of different power sources. Following these studies, this study constructs the CEO power index using the Principal Component Analysis (PCA) based on three power sources: CEO gender, CEO tenure, and CEO ownership. Sheikh (2019) and Harper and Sun (2019) measure CSR implementation by comparing total CSR strengths and real CSR concerns across multiple dimensions. The strengths of CSR represent actions that positively impact social performance, while CSR concerns represent actions that reduce social performance. Unlike previous studies, this study builds a CSR scoring index according to the CSR disclosure approach. The CSR index is based on an annual report scored in four categories built on previous CSR studies, such as environment, community, employees, and products (Muttakin et al., 2018).
Our study generates the following striking results. Firstly, our results show that powerful CEOs reduce the CSRs of commercial banks in Vietnam. Our findings show that when the CEO overpower index increases one score, the CSR index reduces by 0.313. Barnea and Rubin (2010) report that powerful CEOs tend to focus less on CSR investments due to equity ownership in firms. They believe that shareholders' wealth will decline with excessive CSR participation. Powerful CEOs protect their management positions by reducing CSR expenses to empower stakeholder benefits. Therefore, CEOs are less likely to be discharged by the board of directors.
Secondly, our study reports a positive relationship between state ownership and CSR. Specifically, a percentage increase in state ownership increases the CSR index by 0.0314. Lau et al. (2016) and D. Li et al. (2016) suggest that the government provides vital resources and incentives to state-owned enterprises. Firms are motivated to support the government to achieve social objectives in exchange. Therefore, higher state ownership ensures that listed firms improve their CSR. Our findings align with the stakeholder theory and agency theory.
Thirdly, this study documents a moderating role of state ownership because powerful CEOs in state-controlled commercial banks have a higher level of CSR engagement than in private commercial banks. D. Li et al. (2016) suggest that the government creates economic conditions and prioritizes helping state-owned enterprises facing difficulties due to CSR participation. In addition, socio-political orientation is an essential category for government officials to evaluate and consider the promotion level of CEOs in state-owned enterprises. As a result, powerful CEOs in stateowned enterprises are more likely to participate in CSR programs.
We further test whether our findings are robust across CEOs' age subsamples. Our robustness test confirms that our main findings are persistent in all subsamples. Our robustness test results conclude that the older the CEO, the more negative impact on CSR. Although state ownership helps improve CSR, this moderating role cannot outweigh the adverse effects of powerful CEOs on CSR. Interestingly, our findings show that CEOs pay the lowest attention to developing CSR when they are 40 to less than 60 years old. CEOs prioritize developing bank performance rather than improving CSR programs to improve public relationships.
Our study is unique because of the following ways. Our study is similar to Harper and Sun (2019) and Y. Li et al. (2018) because we employ CEO age, ownership, and tenure. However, our study complements prior literature because we use Principal Component Analysis (PCA) to construct the CEO power index. PCA can combine the original variables of CEO power into a one-dimensional index and give consistent results. The results will be multidimensional and heterogeneous if CEO power is distributed according to each power proxy. Specifically, we add the interaction variable between CEO power and state ownership to evaluate the moderating effect of state ownership on the relationship between CEO overpower and CSR. Finally, our results reveal that powerful CEOs in state-owned enterprises are more likely to participate in CSR programs than other banks.
Our study is also similar to Nguyen et al. (2019) because they examine CSR in Vietnam. Our article extends Nguyen et al. (2019) because we further test different CEOs' age subsamples. Gu (2022) argues that misconducts are more serious when CEOs are younger. On the other hand, Altunbaş et al. (2018) show that older and powerful CEOs are more likely to engage in business misconduct, such as bribery and unethical behavior. Therefore, our study generates helpful trends about the impacts of powerful CEOs and state ownership when CEOs are over 30,40,50 and 60 years old. Finally, we figure out that CEOs have less tendency to develop CSR programs when they are from 40 to 60 years old.
The rest of this study is structured as follows. Section 2 provides the literature review. Section 3 describes the research methodology, including data, model formulation, and estimation. The empirical results are reported in section 4. Section 5 is a conclusion.

Agency theory
Based on the agency theory of Jensen and Meckling (1976), Barnea and Rubin (2010) found that the central causal relationship between managers and shareholders is CSR involvement and suggested that managers are interested in CSR investments to promote their private interests. On the other hand, Jo and Harjoto (2011) and Pucheta-Martínez and Gallego-Álvarez (2021) support the conflict resolution perspective, which states that managers engage in CSR to resolve conflicts between different stakeholders. The connection between CEO power and CSR disclosure is a positive sign to maintain the relationship for the benefit of all stakeholders. It is a way for CEOs to reduce stakeholder pressure, reduce their oversight, or improve CEO compensation or tenure. Therefore, when CEOs are powerful, they are expected to be more involved in CSR in this view. Jiraporn and Chintrakarn's (2013) findings are twofold among the sparse literature on CEO overpower and CSR. They report that a positive relationship between powerful CEOs and CSR aligns with agency theory. However, powerful CEOs significantly reduced CSR engagement because CSR investments reduce free cash flow for operations. Therefore, CEOs become entrenched because they have different beliefs on CSR programs. Besides, powerful CEOs do not necessarily satisfy stakeholders through CSR participation because there are difficulties in discharging powerful CEOs.

Stakeholder theory
Stakeholder theory holds that banks and society are interdependent, and therefore, banks should serve a holistic social purpose rather than simply maximizing the interests of shareholders (Rouf & Hossan, 2021). Stakeholder theory suggests managers, especially CEOs, will simultaneously engage in CSR to satisfy different stakeholder interests. Therefore, they integrate social needs into their business model (Freeman & Evan, 1990). Jo and Harjoto (2011) and Pucheta-Martínez and Gallego-Álvarez (2021) argue that CSR involvement can resolve conflicts among stakeholders and thus maximize shareholder wealth. Harper and Sun (2019) argue that powerful CEOs reduce CSR performance because they are less involved in CSR activities to maximize their benefits, such as increasing their authority with stakeholders. Thus, the CEO overpower significantly negatively affects CSR due to agency theory. However, Jiraporn and Chintrakarn (2013) argue that influential executives perceive CSR investments as mechanisms to increase corporate value. Therefore, they tend to invest sensibly in CSR to balance stakeholders' interests.
On the other hand, the government represents the collective interests and promotes the socially responsible behavior of businesses. Xu and Zeng (2016) noted that state-owned enterprises have better CSR performance than other firms because of the government's political and social goals. Stakeholder theory suggests that businesses must win the recognition and support of their stakeholders to continue operating and will tailor their activities to meet stakeholder expectations (Siu & Zhang, 2022). State-owned enterprises are oriented to maximize profits but must also engage in socially responsible activities. Therefore, state ownership is positively correlated with CSR implementation.

Legitimacy theory
Legitimacy theory holds a social agreement between business and society. Institutions strive to meet society's expectations and fulfill social contracts to legitimize existence. For example, banks can legitimize their existence by revealing their actions to curb poverty and unemployment and contribute to education and society (Ullah et al., 2019). Unlike shareholder and agency theories, when only social expectations meet, society will allow the organization to continue operating and ensure its survival. Regarding CSR: A company can legitimize its activities by doing good CSR and winning awards for suitable CSR activities. The gap in legitimacy between business and society will be minimal. Tilling and Tilt (2010) expect that larger companies are more inclined to engage in CSR to create image and value for businesses.

Corporate social responsibility
The concept of corporate social responsibility (CSR) is familiar in the banking industry. However, in the current economic situation, CSR has become a good solution, integrating ethical principles and ethics in banking. CSR has been introduced to Vietnam since the beginning of the 20th century. This market has raised awareness of CSR, recognizing the need. Balancing economic progress, social improvement, and environmental sustainability is essential. The individual level, the organizational context, and the external social context influence the nature of CSR in Vietnam. Nguyen et al. (2019) suggest a relationship between CSR and the development of a country and that CSR initiatives relate to issues of local communities, environment, culture, and economic conditions in society. Suppose a bank is considered socially responsible in its operation. In that case, there are always policies and programs related to social responsibility and developing sustainability, such as taking care of and paying attention to the rights and interests of employee benefits and protecting customers' interests. Thus, implementing CSR at the bank will positively affect society and bank performance. Jiraporn and Chintrakarn (2013) found that CEO power positively influences CSR engagement. Due to increasing their privacy interests, powerful CEOs increase investments in CSR even more. Furthermore, Y. Li et al. (2018) showed a positive association between CEO power in the UK and CSR disclosure. CEOs are encouraged to engage more in CSR for various opportunistic reasons, such as building their public image, marketing strategies, and corporate value enhancement. Barnea and Rubin (2010) reflected that CEO power tends to be less focused on CSR investments due to equity ownership in firms. They believe that shareholders' wealth will decline with excessive CSR participation. Some prior research by Sheikh (2019) and Harper and Sun (2019) investigated the relationship between CEO power and CSR in the US. Consistent with stakeholder theory, they argue that powerful CEOs have an insignificant impact on CSR. Jia et al. (2022) also report that ownership does not affect the CSR program. Powerful CEOs prioritize protecting their positions by maximizing shareholder wealth, so they are unlikely to develop CSR programs.

CSR and CEO overpower
As there are mixed findings between CEO overpower and CSR, we propose the following hypothesis:

State ownership and CSR
State ownership motivates enterprises to pursue social goals rather than profit maximization, implying higher CSR investments. For example, state-owned enterprises are established with economic and social goals and are often subjected to political pressure to please the Government (Zhou et al., 2017). Research by Lau et al. (2016), D. Li et al. (2016), and Chi et al. (2022) show that CSR positively relates to state ownership. The government allocates significant innovation resources and legitimacy to state-owned enterprises. The government often interferes in the annual operations of state-owned enterprises by voting to appoint CEOs for political gain rather than for their competence (Zhou et al., 2017). The extent of government influence is expected to promote the voluntary adoption of CSR commitments and implementation (D. Li et al., 2016). By demonstrating responsible social responsibility and conformance to the expectations of government agencies, state-owned enterprises with power and priority access to critical resources can improve efficiency and profitability.
The Vietnamese government has equitized state-owned commercial banks as part of the privatization process. However, the government maintains sufficient ownership to retain effective control and political influence in public banks. Executives of State-owned commercial banks tend to make decisions and policies for the benefit of the government, such as social and economic benefits.
As prior studies document the mixed findings between state ownership and CSR, we propose the following hypothesis.

The moderating role of state ownership on CEO power
Agency theory supports the role of centralized ownership in the social dimensions of environmental and business regulations. The government, which holds a substantial stake in state-owned enterprises, can reduce agency costs by engaging the management of powerful CEOs in social activities (Jensen & Meckling, 1976). Xu and Zeng (2016) look at companies in China and find that the State as the main shareholder is positively associated with CSR indicators, including governance and social and environmental scores. Therefore, a company with higher state ownership is expected to have a higher quality of CSR. Moreover, the government creates privileges and prioritizes helping state-owned enterprises to motivate their CSR participation because the governments need to archive social objectives rather than maximizing profits. In addition, socio-political orientation is an essential category for government officials to evaluate and consider the promotion of CEOs in state-owned enterprises. As a result, powerful CEOs in state-owned banks will make a more substantial effort to meet government expectations, contributing to improved CSR programs. (D. Li et al., 2016).
On the other hand, state ownership weakens the link between CEO overpower and CSR programs. When the government becomes a significant shareholder, there are conflicting objectives between the government and CEOs. The goals of state-owned enterprises include financial and social objectives simultaneously, while CEOs are more interested in cost-benefit analysis (Jiraporn & Chintrakarn, 2013).
Although Vietnam has transformed from a centralized economy to a market economy, the privatization process is behind expectations. Moreover, the government retains influential roles in the banking sector because the central banks require full support from commercial banks to implement the monetary policies. Therefore, we conjecture that state ownership has a moderating role in motivating CEOs to develop CSR programs in commercial banks. Consequently, we propose the following hypothesis.
Hypothesis 3. Powerful CEOs in banks with state ownership positively affect CSR developments.

Data
We collected data on 38 commercial banks in Vietnam during 2010-2020, excluding joint venture banks and 100% foreign-owned banks. Bank-specific data is manually compiled from various reliable databases, including Vietstock and financial statements of all commercial banks. Agribank is a bank established with 100% capital from the state budget. Therefore, we exclude Agribank because it is a policy bank rather than a commercial bank. To minimize the problem of extreme values, we follow Duong et al. (2021) to redefine all variables at 5% and 95%. We also followed Duong et al. (2022) to remove observations that did not have enough data to calculate relevant variables. Our final sample is a balanced panel with 331 annual observations from 37 commercial banks in Vietnam.

Corporate social responsibility
We attempt to measure CSR using a disclosure approach. In Vietnam, implementing a social index/ ranking approach is impossible because no social rating agency evaluates the social activities of Vietnamese enterprises. Axjonow et al. (2018) also stated that CSR disclosure plays a vital role in social ranking or reputation index. The disclosure approach using content analysis of annual reports is better for Vietnam as it explains the complexity of CSR. We follow Aggarwal and Singh (2019) and classify CSR disclosure into four main areas: environment, society, employees and products, customers, and suppliers. The index ranges from 0 to 4 and reflects CSR disclosure.

CEO overpower
We follow Brodmann et al. (2022) and Altunbaş et al. (2020) to estimate the CEO overpower index by applying the principal component analysis (PCA). PCA eliminates correlated features resulting in independent components. Additionally, the unique function of PCA is to reduce the risk of overload when linearly transforming the high-dimensional variables into a single low-dimensionality index (Ali et al., 2021(Ali et al., , 2022. Specifically, we apply PCA to estimate the CEO overpower index from the three following proxies to realize the research objective. The first proxy is CEO ownership because CEO ownership represents the authority of the CEO, and it significantly influences CSR (Muttakin et al., 2018;Sheikh, 2019). The second proxy is CEO gender because Haider and Fang (2018) suggest that female CEOs overlook risky investments more than male CEOs.
Moreover, Ullah et al. (2019) argue that female CEOs benefit banks because they help reduce risk and promote more prudent and sustainable management. The final proxy is CEO tenure because longer tenure increases CEO power (Sheikh, 2019). Similarly, for three components explained 34.99%, 33.5%, and 31.51% of the overall variations, respectively. Appendix A shows the eigenvectors loadings of PC1, PC2, and PC3. As can be seen in table b, the figures for PC1 and PC2 are small and negative, while PC3 contains significant and positive values. PC 3 explains more data variation, so we collect PC 3 to generate CEO overpower.

Bank size
We follow Ullah et al. (2019) to estimate bank size as measured by the logarithm of total assets. Large and conspicuous enterprises report high-quality CSR to avoid unwanted attention and negative publicity. Xu and Zeng (2016) recognized that firm size positively affects CSR engagement. Large companies may be faced with public concern and forced to invest more in social responsibility. Besides, large enterprises often have more resources, so they are more likely to prioritize CSR.

Board size
Following previous studies by Lau et al. (2016) and Ntim et al. (2013), we also control for the size of the board of directors (BOARD). Consistent with legitimacy theory, larger boards might be more inclined to engage in activities that affect society (Liao et al., 2018). Ntim et al. (2013); Liao et al. (2018) posit that board size is significantly positively associated with CSR participation. They argue that large councils in China representing diverse stakeholders' interests influence a better understanding and grasp of the benefits of ensuring CSR. Rouf and Hossan (2021) observed that CSR was not significantly related to the board size of banks in Bangladesh.

Leverage
Following Boubakri et al. (2016) and Ullah et al. (2019), we also control for the leverage ratio (LEV), which is measured by the proportion of debt in total equity and liabilities. Yang et al. (2018) found a positive relationship between CSR and Leverage. Sheikh (2019) argues that companies with higher CSR are more likely to reduce leverage when market competition is high. Enterprises with CSR use lower debt in their capital structure and are risk-averse, so they are willing to keep a lower leverage ratio.

Model construction
We follow Jiraporn and Chintrakarn (2013) to examine the relationship between CEO overpower and CSR, precisely aligning with personal interests as CSR is enhanced. Engaging in opportunistic behaviors is the primary effect directly from CSR. CEOPOWER represent by CEO gender (CEO_GENDER), CEO ownership (CEO_OWN), CEO tenure (CEO_TENURE), and how to affect CSR, which represents the dependent variable. Our baseline regression model is as follows: (1) Jo and Harjoto (2011) mention that from the Stakeholder theory's point of view, CSR involvement has the potential to resolve conflicts among stakeholders and thus maximize shareholder wealth. On the other hand, the government represents the collective interests and promotes the socially responsible behavior of businesses. Prior studies such as Xu and Zeng (2016), and Zhou et al. (2017) report that state ownership structure has mixed effects on CSR, so we examine how state ownership affects CSR in our model: Our control variables all have an impact on CSR, including LEV (Yang et al., 2018), SIZE (Ullah et al., 2019), and BOARD (Lau et al., 2016). Therefore, we examine how to control variables that affect CSR in both models.
We follow D. Li et al. (2016) to add the interaction variable CEOPOWERxSTATEOWN to the regression model to evaluate the moderating role of state ownership to CEO overpower in the implementation of CSR of commercial banks in Vietnam. The model can be expressed as follows: Where CSR, t represents corporate social responsibility; CEOPOWER, t denotes CEO overpower index; STATEOWN i,t is the percentage of state ownership; Control i,t includes bank size, the board size, and leverage. α i is the firm fixed effect, and α t is the year fixed effect. µ it is the residual value. All variable definitions are displayed in Appendix B.

Estimation procedure
We implement Hausman and Redundant Tests to select the most suitable estimation method among the Pooled Ordinary Least Square (OLS), Fixed Effects Model (FEM), and Random Effects Model (REM). However, Greene (2005) argues that OLS, FEM, and REM violate endogeneity and heterogeneity issues. Because of the limitation of OLS in estimation, there is bias due to variable variance, autocorrelation, or endogeneity issues. With the advantage of using the crosscorrelation equation to improve efficiency and the optimal weight matrix to explain the serial correlation effectively and variable variance, the dynamic system Generalized Method of Moments (GMM) is used to correct all the estimation violations to ensure that the obtained estimates are robust and efficient. Notably, the dynamic system GMM can control endogenous heterogeneity problems, omitted variables measurement error, and autocorrelation (Ali et al., 2021;Shabir et al., 2022;Tran et al., 2022). Therefore, dynamic system GMM provides better and more accurate results.
In addition, the study was carried out on the F-test, AR test, and Sargan/Hansen test to check the suitability of the GMM method. F-test to check the statistical significance of the estimated coefficients. AR test to determine whether the model has a first and second-order autocorrelation. Sargan -Hansen test to check the rationality and appropriateness of the variables selected as instrumental variables. Finally, we perform a robustness test to confirm the persistence of our main findings across subsamples by CEO age. Table 1 presents the descriptive statistics of all variables. The average value of CSR is 3.12, with a standard deviation of 0.88. The average value of CSR in Vietnam is higher than in developed countries like the US (0.188). In addition, the mean of the CEO tenure variable is high in the US, with an average value is 7.61, yet CEO gender and CEO ownership are lower by 0.97 and 0.2, respectively. CEO ownership in Vietnam means 1.05 is higher than in China, with 0.005. The average State ownership in China is 63.15% and in Singapore is 23.50%, much higher than in Vietnam. However, Vietnam has state ownership higher than Hong Kong, the UK, Germany, Japan, and France, which have state ownership at 1.4%, 0.08%, 6.3%, 0.8% 5.11%, respectively. Leverage with a mean of 11.43, SIZE is 18.41, and average board members are 7.38.

Pearson correlation matrix
The correlation relationships between the variables have mixed negative and positive effects. The correlation coefficient of bank size (SIZE) and Leverage (LEV) is about 0.67, showing a strong positive relationship. CEO power and leverage are about 0.04, indicating a weak correlation. We examine the variance inflation factor (VIF) to test whether our sample has a multicollinearity issue. The result is that the VIF of all variables is less than five, and our study has no multicollinearity issue (Tran et al., 2022). Table 3 shows the result of regression by using the dynamic system GMM method. CEO overpower has a significantly negative influence on CSR. Specifically, when the CEO overpower index increases by one score, the CSR index reduces by approximately 0.3130 points. Our finding is consistent with    Table 3 reports the results of dynamic system GMM estimation. Our study examines 37 commercial banks from 2005 to 2020. All variable definitions are reported in Appendix B. The symbol ***, **, and * represents the significant level at 1%, 5%, and 10%, respectively. Barnea and Rubin (2010), implying that powerful CEOs have less attention to developing CSR programs due to agency costs. They believe that shareholders' returns will decline with excessive CSR participation. In addition, our findings are consistent with stakeholder theory, arguing that CEOs overpower an insignificant impact on CSR. They are more likely to take action to maintain or strengthen their power at the cost of stakeholders. Rashid et al. (2020) indicate that CEO power is negatively related to CSR disclosure level because the goal of powerful CEOs is to invest in profitable activities and is not interested in CSR disclosure. Therefore, our finding supports hypothesis 1, implying an inverse relationship between CEO power and CSR. Our result also aligns with stakeholder theory. Table 3 reports that state ownership positively improves CSR programs. Our findings imply that a percentage increase in state ownership improves the CSR index by 0.0314 scores. Our results are consistent with Lau et al. (2016); D. Li et al. (2016) because the degree of government intervention in business performance expects to promote the voluntary adoption of good CSR engagement (D. Li et al., 2016). Besides, our findings are consistent with stakeholder theory. Jiraporn and Chintrakarn (2013) argue that influential executives perceive CSR investments to increase corporate value. In addition, the government represents the collective interests and promotes the socially responsible behavior of businesses. Xu and Zeng (2016) find that state-owned enterprises have better CSR performance than other firms because of the government's political and social goals. Therefore, our findings support hypothesis 2 and stakeholder theory. Table 3 reports that the interaction variable CEOPOWERxSTATEONW positively affects CSR. Specifically, the powerful CEOs in banks with state ownership increase the CSR index by 0.76 percentage points compared with other banks. The results are consistent with D. Li et al. (2016) because the government creates economic conditions and prioritizes helping state-owned enterprises facing difficulties because of CSR participation. Therefore, powerful CEOs in state-owned commercial banks are motivated to implement CSR programs as a development strategy in the banking sector. One of the essential strategies of the banking system is to raise awareness and social responsibility and promote green growth and sustainable development. From the above arguments, there are grounds to conclude that powerful CEOs in banks with higher state ownership are more motivated to develop CSR programs than competitors. Jensen and Meckling (1976) find that high levels of government ownership encourage CEOs to achieve government nonfinancial goals, thereby reducing agency costs. Therefore, our findings support hypothesis 3 and agency theory. Table 3 reports that bank size (SIZE) has a significant positive effect on the models. CSR of banks tends to increase by 0.3211 scores if the log of total assets increases by 1 unit. Our findings, consistent with Xu and Zeng (2016), recognized that firm size positively affects CSR engagement. Large companies may be exposed to public pressure and forced to invest more in social responsibility to develop their public image as a public relation strategy. Table 3 indicates that Board size (BOARD) negatively impacts CSR. Specifically, an additional board member reduces CSR by 0.053 scores. Banks with large boards are less likely to engage in social responsibility. Our findings agree with D. Li et al. (2016), who suggested that board size is negatively associated with CSR participation.

Regression results
Finally, Table 3 suggests that Leverage (LEV) discourages CSR. The coefficient of LEV is −0.0002, implying that CSR reduces the 0.02 scores if the leverage ratio increases by 10%. Sheikh (2019) argues that higher leverage reduces CSR when market competition is high. A higher leverage ratio implies higher interest payments, reducing the resources available to develop CSR programs. Gu (2022) argues that misconduct is more serious when CEOs are younger. On the other hand, Altunbaş et al. (2018) show that older and powerful CEOs are more likely to engage in business misconduct, such as bribery and unethical behavior. We conjecture that misconducted CEOs are unlikely to develop CSR programs. Therefore, we examine whether this section's tendency to participate in CSR programs changes with CEOs' age. Table 4 reports the negative impacts of powerful CEOs on CSR across CEO age subsamples. Harper and Sun (2019) suggest that powerful CEOs are less involved in CSR because they focus on increasing their power with stakeholders. Barnea and Rubin (2010) reflected that CEO overpower tends to be less focused on CSR investments due to equity ownership in firms. They believe that shareholders' returns will decline with excessive CSR developments. In addition, Table 4 indicates that higher state ownership positively contributes to CSR development across CEO's age subsamples. Xu and Zeng (2016) found that state-owned enterprises have better CSR performance than other firms because of pursuing the political and social goals of the government. Therefore, Table 4 reports the robust impacts of powerful CEOs and state ownership on CSR.

Robustness test
Finally, we figure out the decreasing trends of CSR developments when powerful CEOs get older. For every score increase in CEO overpowers index, the CSR index reduces by 0.1832, 0.2185, 0.4261, and 0.3037 across subsamples by size. Our findings imply that CEOs in the early stage tend to develop CSR the most before 40. However, CEOs have the lowest attention to improving CSR from 40 to less than 60 years old.

Discussion
This study explores the impact of powerful CEOs and state ownership on effective CSR programs in Vietnamese banks. The findings of this study complement CSR studies, especially in the banking sector. Our findings indicate a negative relationship between CEO power and CSR. Our evidence suggests that more powerful CEOs are less interested in CSR programs because investments in CSR are wasteful. In addition, if these CSR activities reduce stakeholders' interests, powerful CEOs limit  Table 4 reports the CSR of commercial banks with a CEO over 30-60 years old using the system GMM method. All variable definitions are reported in Appendix B. The symbol ***, **, and * represents the significant level at 1%, 5%, and 10%, respectively. the implementation of CSR programs to ensure benefits and maintain relationships with stakeholders. It could be a way for CEOs to relieve pressure, reduce their oversight, or improve CEO power (Pucheta-Martínez & Gallego-Álvarez, 2021). Thus, hypothesis 1 is accepted. Unfortunately, our findings are inconsistent with Jiraporn and Chintrakarn (2013); Y. Li et al. (2018).
In addition, our study shows that state ownership positively increases CSR programs. Xu and Zeng (2016) find that state-owned enterprises have better CSR performance than other firms because of the government's political and social goals. In addition, state-owned enterprises actively participate in CSR to show that the government acts responsibly toward the public (Siu & Zhang, 2022). Our findings align with Lau et al. (2016), D. Li et al. (2016), and stakeholder theory. Our results also support hypothesis 2.
Moreover, the interaction variable between CEO power and state ownership positively affects CSR, indicating a moderating role of state ownership in empowering the effectiveness of CSR programs. D. Li et al. (2016) analyzed that the government creates economic conditions and prioritizes helping state-owned enterprises facing difficulties because of CSR participation. Powerful CEOs in banks with higher state ownership are more motivated to develop CSR programs than in private banks. Therefore, our findings also support hypothesis 3.
We further extended the research results when testing robustness across CEOs' age subsamples. Our robustness test confirms that our main findings are persistent in all subsamples. Interestingly, our findings show that the older the CEO, the lower perception of CSR. CEOs aged 40 to less than 60 prioritize developing bank performance rather than improving CSR programs to improve public relationships and brand image.

Conclusion
This study examines the extent of CSR implementation and disclosure in Vietnamese commercial banks and the impact of CEO overpower on CSR implementation. In addition, it also demonstrates the moderating role of state ownership and CEO overpower in the banking and financial sector. We collected a sample of 37 commercial banks from 2010 to 2020. We employ the dynamic system GMM to analyze data and discuss our findings.
Empirical results show that powerful CEOs negatively reduce CSR. CEO power tends to be less focused on CSR investments due to equity ownership in firms. Meanwhile, our findings indicate a positive relationship between state ownership and CSR, implying that higher state ownership enhances CSR significantly. However, the interaction variable between state ownership and CEO power positively affects CSR, suggesting a moderating role of state ownership. Powerful CEOs in banks with higher state ownership have more motivation to develop CSR programs than competitors. Our findings are consistent with the stakeholder theory, Lau et al. (2016), Xu and Zeng (2016), D. Li et al. (2016), Barnea and Rubin (2010), and Rashid et al. (2020). Our primary findings remain robust when we test different CEOs' age subsamples. Finally, CEOs in the early stage tend to develop CSR the most before they are 40 years old. However, they have the lowest attention to improving CSR from 40 to less than 60 years old.
Our study contributes important practical implications for academics, investors, and market agents. Firstly, this study suggests policymakers pay more attention to corporate governance mechanisms to monitor the CEO's power to promote CSR programs. In addition, market agents should consider maintaining the moderating role of state ownership in the banking system to improve the effectiveness of CSR implementation.
Secondly, the findings assist bank managers in developing the effectiveness of CSR programs. Investors may be more interested in state-owned enterprises where powerful CEOs carry out CSRrelated activities that increase shareholder returns. Investors can also benefit from these strategic decisions in the form of improved returns and demonstrate their commitment to social and environmental issues to society and stakeholders.
Finally, this study contributes to theoretical implications because it provides an integrated theoretical framework and empirical evidence of how powerful CEOs and state ownership affect CSR programs. This study may serve as a guideline for students and academics to extend the corporate governance and CSR topics. From a theoretical point of view, our findings suggest that CEOs who become more powerful are more likely to make decisions without regard to the interests of their stakeholders, which can lead to a low level of social responsibility and, therefore, affect the implementation of CSR activities.
This study has the following limitations. Firstly, it has data limitations because it only reflects the context in Vietnam. Second, Vietnam is a frontier market, so our findings may not reflect the contexts in emerging and developed markets. Therefore, we suggest that future studies examine this topic across countries to generate in-depth insights.