Do Corporate Social Responsibilities Support Its Financial Performance? Evidence of the Listed Firms in Vietnam

: In the latest decades, corporate social responsibilities (CSR) are incrementally noticed in emerging countries regarding to the surge of globalization and the advance of social concerns. However, there exist opposing arguments on how CSR practices influence a firm’s financial performance (CFP), either in theoretical or empirical perspectives, causing many managers’ hesitation in CSR engagement. This study, therefore, examines the impacts of CSR on CFP, considering the individual effect of each CSR endeavor by using the data from listed firms’ financial statements during the period 2015 -2019 and panel regression analysis methods. The findings reveal inconstant effects of different CSR activities on CFP. Specifically, a firm’s fulfillment of shareholders’ and cus tomers’ interests contributes to raise its profitability, while its engagement in the benefits of employees and creditors causes a reduction in financial returns. Interestingly, the correlations between firms’ responsibilies towards regulators and suppliers and CFP are statistically insignificant. Furthermore, we make pairwise marginal comparisons to identify the distinctions of CSR-CFP relations across industries. The results only indicate that listed firms in Utilities sector have the lower level of CSR intensity than those in Consumer Discretionary and Consumer Staples sectors. The implications and limitations are also discussed in this study. The regression models demonstrate inconsistent impacts of different CSR practices, as well as corporate characteristics, on CFP. First ly, WAGE, representing the firm’s responsibility towards employees, is negatively correlated to the firm’s financial performance in all three models at 1% significance level. Besides, the firm’s assurance of interest expense (ICR) has a negative nexus with CFP at 5% significance level in pooled OLS model, and 10% in REM. These results are consistent with neoclassical views, which regard social performance as the cause of excessive costs and resource misallocations that may harm the firm’s profits and its owners’ wealth (Friedman, 1970; Lyon et al., 2013; et Nollet et al., the ratio of pay-out dividends to common shares has a significantly positive linkage with CFP, at the significance level of 1% in pooled OLS and FE model, and 5% in REM. Also, the coefficient of OEx is 0.00002 in the robust RE model, which indicates a statistically, but faintly, positive correlation between a firm’s concerns to customers and its profitability at the 1% significance level. These results support two parallel sides: a firm’s fulfillment of shareholders’ interests can attract their efficient participations into internal decision- making procedure, while its responses to consumers’ concerns contr ibute to raise their satisfaction and confidence that fosters the firm’s reputation, intangible value, and future value Dmytriyev, 2017; Tseng the study to Okafor, Adusei, and Adeleye, (2021) ’s a firm’s CSR are not automatically converted into but rely on what and how those are perceived consumers’ asses sment, firms’


Introduction
Since the early age of business, profit maximization for the owners, then shareholders, is the mere goal of every firm, while actions towards nonowners' benefits is primely considered as a "doctrine" or public deception (Friedman, 1970). However, followed by the rapid change in the global business environment and the social awareness of responsible investment and consumption, corporate social responsibilities (CSR) is raised as the new trend of business strategies (Carroll, 2015;Low, 2016). These activities are believed not only to fulfill the stakeholders' expectations but also to help firms gain certain advantages (Freeman & Dmytriyev, 2017). Nonetheless, academic arguments over the relation between CSR and corporate financial performance (CFP) are still inconsistent (Orlitzky, Schmidt, & Rynes, 2003), causing senior managers' hesitations to activate their social schemes (Carroll, 2015).
Since the last decades, CSR are officially implemented and disclosed in developed economies (Belal, 2001). Then, the globalization progress and the extension of multinational enterprises has encouraged the implementation and standardization of socially responsible activities in less developed markets. However, in these countries in general and Vietnam in particular, the stakeholders' interest of CSR activities is still fragmented; the question of whether firms can become socially responsible, and whether these responsibilities create a positive spillover to their performance in emerging economies remains unreciprocated (Birch & Moon, 2004;Jamali & Karam, 2016). On the one hand, this discrepancy is arisen from many differences in institutional, cultural and socio-economic backgrounds (Al-Mamun & Seamer, 2021;Auer & Schuhmacher, 2016). On the other hand, the majority of previous studies merely refer to CSR as a single measure without intensive considerations (Busch & Friede, 2018). Several authors tend to use available indexes aggregated by third-party organizations, e.g., Toxics Release Inventory (TRI) by EPA (US), ESG scores by Thomson Reuters or Bloomberg, or KLD 400 Social index by MSCI, which cover a minor portion of corporates in developing economies (Belal, 2001;Ho & Yekini, 2014;Ngoc, 2018). Accordingly, the economic theories and models, research findings and policy implications emerged from developed countries may be not well compactible to other countries (Sardana, Gupta, Kumar, & Terziovski, 2020). Besides, the others have focused on a few of CSR aspects, e.g., Ausat (2018) examines the influence of some corporate governance variables on Islamic banks' financial performance, Ho and Yekini (2014) investigate the bidirectional relation between CSR disclosure and CFP of 20 largest financial firms in Vietnam, Ngoc (2018) finds the effect of CSR disclosure on Credit institutions' financial performance in Vietnam. To our best knowledge, no previous studies has focused on the thorough relationship between individual CSR variables and listed firms' financial performance in case of Vietnam.
Hence, to provide an empirical framework for the corporates' making-decision process in Vietnam, this study makes a deep investigation on the correlation between CSR practice Be desired to be a good Corporate-citizen Be expected to be ethical Be required to obey the law Be required to be profitable and CFP. The study uses the financial data of listed companies on Ho Chi Minh Stock Exchange (HSX) during the period between 2016 and 2019 to consider the individual influences of distinctive CSR activities on CFP, and examines the variety of this relation across different industries. Subsequently, the paper makes great contributions by identifying the insightful influences of various CSR practices on CFP in the specific context of Vietnam, which does not exist in previous studies. The remaining of study is structured as follows: section 2 reviews the theoretical backgrounds, previous studies on the CSR-CFP relation, and develops hypotheses. Section 3 and 4 exhibit the research methods, empirical results, and discussions. Finally, the conclusions and limitations are presented in section 5.

Corporate Social Responsibilities (CSR) Concept
In contemporary society, as the business environment evolves with globalization, advanced technology, and customers' attention to ethical product standards beyond the quality of goods and services, CSR has integrated into business activities as a multidimensional disciplinary subject (Low, 2016). Although the CSR concept has emerged since 1950s and gone globally now, it lacks an acceptable and unique, or official, definition (Rahman, 2011). According to many scholars, e.g., Carroll (1995Carroll ( , 2015, Jamali and Karam (2016), Rahman (2011), CSR's interpretations and expressions vary across the local context of social, economic, political, environmental issues and time periods. In the very first period, CSR primarily consists of a firm's economic contributions such as that of employment, payrolls , and social audits (Carroll, 1995). Then, it is developed into a comprehensive concept that embraces not only economic obligations but also legal, ethical, and philanthropic responsibilities (Carroll, 1979), and well known as the "CSR pyramid" model (see Figure 1). Basically, the firm's social responsibilities involve the simultaneous fulfillment of the society's entire expectations of economic, legal, ethical, and discretionary/philanthropic issues at a certain point of time (Carroll, 1995).

Figure 1
The "CSR pyramid" Source: aggregated from the study of Carroll (1979Carroll ( , 1995Carroll ( , & 2015  Subsequently, modern firms are held responsible for the consequences of their actions, and under pressure to integrate stakeholders' values and concerns into their business strategies that ultimately determine their performance (Carroll, 2015;Gond, El Akremi, Igalens, & Swaen, 2018;Porter & Kramer, 2011). Since Freeman's book "Strategic Management: A stakeholder approach" in 1984, stakeholders are defined as any subjects relevant to a firm's target-reaching process, typically including employees, suppliers, financiers (owners, investors), customers, communities, and regulators; subsequently, stakeholders' active participations are regarded as the major component of firms' CSR engagement, and simultaneously, an integral part of firms' performance, which makes the stakeholder theory the most essential and widely used foundation (Rahman, 2011).
The CSR-CFP relationship and hypothesis development The investigation of the potential relationship between CSR and CFP remains its attractiveness globally over fifty years, creating a long record of empirical studies based on various theoretical frameworks (Gond et al., 2018). Yet, the conclusions of CSR-CFP relation remain obscure and conflicting (Friede et al., 2015;López-Arceiz, Bellostas, & Rivera, 2017).
On the one hand, the stakeholder theory basically supports that the firm's overall responsibilities are to enhance relationships and create value for all of its stakeholders (Freeman & Dmytriyev, 2017). Along with the development of business environment and complimentary concepts, e.g., business ethics, corporate citizenship, stakeholder management, sustainability, or creating shared value (Carroll, 2015;Porter & Kramer, 2011), the traditional stakeholder theory has been decomposed into alternative aspects -normative, descriptive, and instrumental perspectives. According to Donaldson and Preston (1995), although all of these are nested within each other, the normative aspect is the central core of the stakeholder theory, suggesting that firms should treat all of stakeholders equally regardless of their prominence. This perspective explains why stakeholder management may casually relate to corporate performance, encouraging managers to take care of all stakeholders' concerns instead of the firm's owners solely. However, due to limited corporate resources, managers find it difficult to implement in practice, especially if stakeholders' interests are not consonant (Fernando & Lawrence, 2014). Conversely, the descriptive, or managerial, perspective claims that firm managers should spend more attempts to accomplish the expectations of prioritized stakeholders who provide and control critical resources. Yet, the key issue of this perspective is how to classify stakeholder groups and identify how much interest the firm's managers should response (Fernando & Lawrence, 2014).
The instrumental perspective is more balancing with the assertion that the satisfaction of all stakeholders is a useful means for firms to achieve their business objectives. Specifically, a firm's performance and value rely on how its managers deal with operating costs, not only arrived from explicit contracts such as wages and investors' earnings but also implicit accords such as that of product/service quality as well (Cochran & Wood, 1984;Mcguire et al., 1988). This argument implies that once the benefits of any stakeholders who have implicit compromises with the firm are damaged, they possibly try to transfer implicit terms into explicit agreements which are usually more costly. Hence, the instrument aspect of stakeholder theory can explain how the firm's capacity of balancing the interests of all stakeholder parties creates competitive advantages compared to its rivals and leads to higher financial performance (Li et al., 2017;Orlitzky et al., 2003). Consequently, it is more commonly applied in literature (Aras et al., 2010;Prasetya, Saraswati, & Ghofar, 2017;Fernando & Lawrence, 2014;Madueño et al., 2016;Mcwilliams & Siegel, 2000;Salehi, DashtBayaz, & Khorashadizadeh, 2018).
On the other hand, neoclassical views claim the negative influence of CSR on CFP. Friedman (1970) denotes that a firm's engagement in social commitments involves both financial and non-financial resources (e.g., human resource) that can be used for investment purposes. Also, the increase of social activities (e.g., charity, environmental campaigns, community development) can cause additional expenses, and lead the firm's managers to failed resource allocations, which push the firm to an economic disadvantage, e.g., a trade-off situation, compared to less socially active firms, negatively impacting the firm's profits and owners' benefits (Preston & O'Bannon, 1997;Waddock & Graves, 1997). Although CSR can be used to prevent or diminish negative impacts of a firm's operations, profit maximization is still the firm's vital target, and CSR practice intentionally reduces the profits that must be available to other stakeholders such as its owners and investors (López-Arceiz et al., 2017;Lyon, Lu, Shi, & Yin, 2013). Friedman (1970) also emphasizes that it is wrong for managers to exercise social responsibilities using the firm's resources, because these are other people's money and must have been spent on customer service, employees' welfares or shareholders' dividends. This mistake may cause the firm to lose fundamental supports to enhance its financial performance, and incurred expenses such as agency costs possibly outweighs the benefits from CSR practice (Mcwilliams & Siegel, 2000). Moreover, the reflection of a firm's profitability in CSR disclosures is restricted (Gallardo-Vázquez, Gallardo-Vázquez, Barroso-Méndez, Pajuelo-Moreno,& Sánchez-Meca, 2019). In practice, CSR aspects are quite subjective and relied on stakeholders' cognitions, thus, information may be confusing, biased, and actually manipulated by a firm's managers to deceive its stakeholders (Kim & Lyon, 2015;Zhao et al., 2018).
Besides, some studies find mixed CSR-CFP relations. Specifically, Mcguire et al. (1988), Dobre et al. (2015) and Yoon and Chung (2018) find the nonlinear relations, Nollet, Filis, and Mitrokostas (2016) find both linear and non-linear relations, while Aras et al. (2010), Soana (2011) and Han et al. (2016) indicate an insignificant correlation between them. According to many scholars, these conflicting findings are caused by different methodologies, such as different measures (e.g., accounting-based or market-based indicators, or even adjusted risk) (Waddock & Graves, 1997); dissimilar identifications of stakeholders, perspectives, time periods, sample size, and selected variables (Dobre et al., 2015;Mcwilliams & Siegel, 2000;Su, Liu, & Teng, 2020); divergent analysis methods (e.g., content analysis, regression models, event studies) (Crifo, Diaye, & Oueghlissi, 2017;Mcguire et al., 1988;Taliento, Favino & Netti, 2019;Zhao et al., 2018). Moreover, the majority of previous studies that have proved the positive relation CSR and CFP focuses on developed countries (Belal, 2001;Friede et al., 2015), in which the ethical, cultural, and business norms, as well as stakeholders' cognitions and interests, are far apart from emerging countries (Birch & Moon, 2004;Carroll, 2015;Garcia, Mendes-Da-Silva, & Orsato, 2017;Jamali & Karam, 2016). Hence, we doubt about this conclusion in the context of emerging countries for two reasons. Firstly, corporates operating in developing markets do not have many options for financing, partly because of the immature capital market in these countries, which may generate an obvious trade-off between CSR and other lucrative activities (Garcia et al., 2017). Secondly, along with economic issues, institutional and cultural background of a country noticeably influence the nature and degree of CSR practices (Jamali & Karam, 2016). Subsequently, we do not expect CSR practices to enhance the firms' profitability operating in Vietnam. The first hypothesis is built as following: H 1 : There is a significantly negative effect of the firm's social responsibilities on financial performance in Vietnam.
To capture the individual influence of each CSR practice on CFP, the following hypotheses are put forward as constituents of H 1 : Furthermore, the samples in these studies include various sectors, while CSR activities vary remarkably across industries and markets (Soana, 2011;Su et al., 2020). Particularly, among numerous firm-specific indicators used as the moderating factor for CSR-CFP relation in empirical studies, e.g., firm's size, leverage, liquidity (Delmas, Nairn-Birch, & Lim, 2015;Dobre et al., 2015), R&D intensity, corporate ownership (Aras et al., 2010;Mcwilliams & Siegel, 2000), financial risks (Mcguire et al., 1988), growth opportunities (Adair & Adaskou, 2015), operating industry, visibility, reputation, income diversification, business cycle… (Busch & Friede, 2018), it is important to take industry categories into consideration, because costs and benefits relating to CSR performance in different industries may be noticeably distinguished from each other (Dobre et al., 2015;Lin, Chang, & Dang, 2015). For example, firms in sensitive industries such as mining, petrol, pulp and paper, steel, chemical, and pharmaceutical production are more likely to perform superior CSR practice, e.g., on employees' health and safety, disclosures of their pollution control, compared to other non-sensitive industries, to avoid moral debates and political pressures (Albertini, 2013;Garcia et al., 2017). Therefore, this study posits the second hypothesis: The effects of CSR on CFP are dissimilar between different industries.

Research Method
Data This paper investigates the relationship between CSR and CFP, considering the moderating effect of industry. Also, given that CSR is a multi-facet concept, and the effects of one aspect sometimes deactivate opposing effects of another aspect (Nollet et al., 2016), the analysis should contain disaggregated proxies. However, previous studies in Vietnam, in our best knowledge, have considered CSR as a single variable. For example, in the studies of Ho and Yekini (2014) and Ngoc (2018), corporates' CSR practices are merely represented by "CSR disclosure", obtained from their annual reports via content analysis, which cannot reflect various dimensions of this concept. Thus, this study decomposes the general CSR concept into the firm's responsibilities towards the regulators, employees, suppliers, shareholders, creditors, and customers, as suggested by Donaldson and Preston (1995).
Data is extracted from the annual financial statements of the non-financial firms listed on Ho Chi Minh Stock Exchange (HSX) during the period between 2015 and 2019. Although there are three official stock exchanges in Vietnam, including HSX, HNX (Hanoi Stock Exchange), and UPCoM (Unlisted Public Company Market), the HSX has higher standardized regulations for listing, which makes disclosed information more available and credible. The study doesn't take financial firms into account due to their distinct characteristics, as well as all firms' performance since 2020 to exclude potential impacts of the Covid-19 pandemic. Additionally, the listed firms are filtered by the completeness of disclosed information, then categorized by industry based on GICS® (Global Industry Classification Standards), a standard for industry classification developed by MSCI and S&P Dow Jones Indexes, and applied on HSX since January 25 th 2016. After filtering, the number of remaining listed firms is 162, which is shown in Table 1. y it = α + β 1 TAX it + β 2 WAGE it + β 3 APT it + β 4 DIV it + β 5 ICR it + β 6 OEx it + β 7 SIZE it + β 8 LEV it + ε it for i = 1, 2, …, n, and t = from 1 to 5

Minh, Hong, & Anh
Where y it is the measure for financial performance of the firm i in the fiscal year t. TAX it , WAGE it , APT it , DIV it , ICR it , OEx it respectively represent the firm's social responsibilities towards regulators, employees, suppliers, shareholders, creditors, and customers. SIZE it and LEV it are control variables that have been justified by prior studies. The description of each variable is shown below: Various indicators have been used as proxies for corporate performance in empirical quantitative studies, mainly subdivided into accounting-based (e.g., ROA, ROE) or marketbased indicators (e.g., Tobin's q, stock returns, sales growth) (Albertini, 2013;Aras et al., 2010;Bergmann, 2016). Although each group of measures have different advantages and drawbacks, the accounting-based indicators, calculated from financial statements, can represent a firm's financial performance and internal power more precisely (Cochran & Wood, 1984), and have stronger linkage to CSR practice than the others (Busch & Friede, 2018;Orlitzky et al., 2003). Thus, this study uses ROA (Returns on Assets) as the proxy for CFP.
Tax intensity (TAX) is calculated by the ratio of the current corporate income tax expense to the earnings before interests and taxes (EBIT) in the fiscal year t. It is noticeable that, according to Vietnamese Accounting Standards (VAS), the current corporate income tax expense is not merely set by the fixed corporate income tax rate (20% in particular), but also adjusted by individual state's preferential policies and sanction regulations. Thus, this amount can reflect a corporate's compliance level to state-related obligations and its relations with municipalities. The municipal services are primely relied on businessrelated taxes, thus, when the public budget is ensured, the government can provide better public services, e.g., improved infrastructure, for business development.
Simultaneously, the firm's compliance to regulations can enhance its legitimacy and reputation, contributing to reduce idiosyncratic risks and attract responsible investors (Garcia et al., 2017).
Wage ratio (WAGE) is computed by the ratio of the wages amount paid to a firm's employees to EBIT in the fiscal year t. Human resource is one of the most critical factors of firms' sustainable development (Ruiz-Pérez, Lleo, & Ormazabal, 2021). The improvement of corporate welfares, directly on wages and salaries, not only enhances the current employees' productivity and loyalty, but also attracts high-qualified personnel with improved internal making-decision process (Salehi et al., 2018;Tseng et al., 2020).
Accounts payable turnover (APT) is measured by the ratio of total supply purchases to the sum of average accounts payable and average accrued liabilities in the fiscal year t, as suggested by Ilter (2020). This measure reflects a firm's short-term liquidity, which is not only beneficial for daily business activities through discounts for early payment, but also useful to develop long-lasting relationships between the firm and its suppliers. Also, Chen, Karim, and Tao (2021) indicate that the integration of suppliers' concerns into a firm's CSR strategies can effectively lessen its financial risk such as stock price crash.
Dividends per share (DIV) is identified as the total dividends paid out to common shares divided by the average number of outstanding common shares in the fiscal year t. On the one hand, this ratio represents the firm's primary target -benefits maximization for current shareholders -to encourage shareholders' active participation and loyalty as well (Lyon et al., 2013). On the other hand, it is regarded as a signal of the firm's growth capacity to attract potential investors, ensuring the sustainable development (Kanakriyah, 2020).
Interest coverage ratio (ICR) is calculated as EBIT divided by the total interest expense in the fiscal year t. Basically, this ratio demonstrates how well a firm can pay interest due on its outstanding debt. The higher ratio reflects the firm's enhanced financial assurance, enhancing its credibility and pleasant relationships with external financiers that possibly reduce its capital expenditures in the future (Salehi et al., 2018).
Operating expenses ratio (OEx) is measured by the ratio of operating expenses to net operating returns in the fiscal year t. According to VAS, a corporate's operating expenses refer to selling, general and administrative expenses (SG&A) extracted from the business results report, which comprise costs directly or indirectly costs related to sales activities, e.g., advertising, marketing, promotion, distributing costs, customer services, as well as wages and benefits for salespeople, administrators, and management staffs who are not involved in manufacturing and other production tasks; net operating returns is calculated by subtracting financial returns and operating expenses from the gross returns of sales. Offering more benefits for consumers delivers higher operating expenses (Mcguire et al., 1988), yet customers' confidence can help the firm to improve its reputation and expand its market share, which contribute to increase its revenues that effectively offset additional costs (Salehi et al., 2018). Similarly, Seetharaman, Moorthy, Saravanan, and

Minh, Hong, & Anh Do Corporate Social Responsibilities Support Its Financial Performance? …
Jurnal Ekonomi & Studi Pembangunan, 2022 | 10 Pitta (2016) prove that SG&A expenditure has strongly positive impacts on sales revenue, gross margins and profits.
Firm size (SIZE) is the common control variable used in empirical studies relating to corporate financial management, and estimated by the natural logarithm of total assets at the end of the fiscal year t, as suggested by Salehi et al. (2018) and Kanakriyah (2020). Larger firms are more apparent to the society than smaller firms, thus, tend to disclose more information regarding socially responsible behaviors (Aras et al., 2010;Waddock & Graves, 1997). Also, they are believed to have less default risks than smaller firms due to the capacity of income diversification, lower transaction and disclosing costs, which are favorable conditions to increase their financial performance (Adair & Adaskou, 2015).
Financial leverage (LEV) is represented by the ratio of the firm's total liabilities to total assets in the fiscal year t. On the one hand, external financing helps the firm to raise adequate capital amount for investment to boost its returns. On the other hand, high leverage implies higher unsystematic risks in the future, e.g., liquidity or default risk, which firm managers must take into considerations during decision-making process (Aras et al., 2010).
This study also examines potential differences in the nexus between CSR and CFP across sectors. Several empirical evidence claim that the estimation model of the CSR-CFP relation may be mis-specified without the industry, because firms' operating costs and benefits relevant to CSR practice may be remarkably distinctive among different sectors (Dobre et al., 2015). Thus, a comparison by industry should be included in this analysis.
ε it is the error term that contains both idiosyncratic error term u it and unobserved firmspecific characteristics σ it , such that ε it = u it + σ it . Depending on the assumption on the unobserved firm-specific characteristics term, σ it , we can run two different linear regressions with the panel data. Under the assumption that σ it varies across time, we run a Random Effects model (REM). In case that σ it does not change across time, we run a Fixed Effects model (FEM). No correlations are found between independent variables and the error term to exclude endogenous problems (Garcia et al., 2017). We also run the pooled Ordinary Least Squares (OLS) regression, taking into account the specific time effect, because the pooled OLS does not distinguish a firm's characteristics and its time trend (Nollet et al., 2016).

Result and Discussion
Before performing a multivariate analysis to test the hypotheses afore posited, the study provides the descriptive statistics in Error! Reference source not found.. In the panel A, the overall average values, and standard deviations of all six variables -TAX, WAGE, APT, DIV, ICR, and OEx -suggest a significant discrepancy of social activities between firms in the sample. For example, firms' responsibility towards the government through tax intensity (TAX) can vary between -0.05 (the firm doesn't have to pay corporate income tax in a fiscal year) to 0.88, with the average value of 0.15; firms' responsibility towards their shareholders through cash dividends (DIV), with the mean at -847.05, ranges from VND -36,271.64 per share up to VND 16,420.51 per share. However, when it comes to the comparison by industry (panel B), statistic values of every single variable (means and standard deviations) are somehow consistent across sectors. Thus, the study expects that the variation of CSR practices happens among listed firms within the same activity sector, not across different industries.
In the matrix of the bivariate correlations, it is expected that the relationship between CSR practice and CFP is quite mixed. Specifically, most of a firm's socially responsible actions, TAX, APT, DIV, ICR and OEx, have positive effects on CFP, with positive correlation coefficients. WAGE is the only variable that create negative influence on financial performance. The bivariate coefficients between CFP and firm-specific variables, SIZE and LEV, are also negative, suggesting a negative linkage between them during the sample period.
The impact of chosen independent variables on CFP is estimated by three different methods (OLS, FEM, REM), which is shown in Error! Reference source not found.. To identify the estimation method suitable for the regression model, the study uses the Hausman's and Breusch & Pagan's test. We find no evidence of a correlation between independent variables and the error term, demonstrating no problems of endogeneity. Specifically, Hausman's test (p > 0.1) accept the null hypothesis that the unobserved heterogeneity is uncorrelated with the independent variables, implying that the model of factors obtained from random-effects regression is consistent; Breusch and Pagan Lagrangian multiplier test (p < 0.01) rejects the adjustment of pooled OLS, suggesting the suitability of REM. Therefore, in this study, the regression model controlling for random effects gives more consistent parameters, indicating the random effects model is the better option.     The regression models demonstrate inconsistent impacts of different CSR practices, as well as corporate characteristics, on CFP. Firstly, WAGE, representing the firm's responsibility towards employees, is negatively correlated to the firm's financial performance in all three models at 1% significance level. Besides, the firm's assurance of interest expense (ICR) has a negative nexus with CFP at 5% significance level in pooled OLS model, and 10% in REM. These results are consistent with neoclassical views, which regard social performance as the cause of excessive costs and resource misallocations that may harm the firm's profits and its owners' wealth (Friedman, 1970;Lyon et al., 2013;Mcguire et al., 1988;Nollet et al., 2016).
Conversely, the ratio of pay-out dividends to common shares (DIV) has a significantly positive linkage with CFP, at the significance level of 1% in pooled OLS and FE model, and 5% in REM. Also, the coefficient of OEx is 0.00002 in the robust RE model, which indicates a statistically, but faintly, positive correlation between a firm's concerns to customers and its profitability at the 1% significance level. These results support two parallel sides: a firm's fulfillment of shareholders' interests can attract their efficient participations into internal decision-making procedure, while its responses to consumers' concerns contribute to raise their satisfaction and confidence that fosters the firm's reputation, intangible value, and future value (Freeman & Dmytriyev, 2017;Madueño et al., 2016;Tseng et al., 2020). However, the study is coherent to Okafor, Adusei, and Adeleye, (2021)'s notice that a firm's CSR actions are not automatically converted into economic benefits but rely on what and how those are perceived by consumers' assessment, thus, firms' efforts to earn customers' pleasure may not result in the corresponding effects.
Secondly, it is surprising that we cannot find any statistically significant effects of either TAX or APT on CFP, with p-value > 0.1 in all three regression models. On the one hand, it makes sense that the connection between tax-related obligations and CFP is trivial because, according to Carroll (1995), the society allows firms to play their economic role -producing goods and services, and earning profits -and in turns, lays down certain ground rules -laws -that firms are required to comply. In other words, the firm's duty to regulators through taxes is somehow the compulsory action to gain legitimacy, a critical factor for its survival and development (Prasetya et al., 2017). On the other hand, this study's result is opposed to the evidence by Chen et al. (2021), refuses any effects of a firm's responsibility to suppliers on CFP. As far as we know, this relation in the specific case of Vietnam has not been tested in previous empirical studies, thus, can be regarded as the interesting contribution of this study.
Thirdly, the study also finds dissimilar influences of different firm-specific factors on CFP. Specifically, the firm's financial leverage (LEV) negatively affects its financial performance in all three models at 1% significant level, consistent with the findings of Kanakriyah (2020), as a trade-off between profitability and risks referred by Aras et al. (2010).
Meanwhile, the firm's size creates a positive linkage with CFP at the significance level 1% in pooled OSL and FE model, and 5% in RE model, which is coherent to the arguments of Quyen et al (2021) Adair and Adaskou (2015), Aras et al. (2010) and Waddock and Graves (1997).
Furthermore, this study also examines the impact of industry categories on the CSR-CFP relation through pairwise comparisons on the marginal estimations, shown in Error! Reference source not found.. The results show that most of p-values of pairwise comparisons are higher than 0.1. This means the distinction of CSR intensity between activity sectors is statistically insignificant. However, there are two exceptions: the difference between group 8 and 1, with the coefficient of -0.031 and p-value = 0.033, stating that firms in group 8 (Utilities) have lower intensity of CSR practices than firms in group 1 (Consumer Discretionary) at 5% significance level; and the difference between group 8 and 2, with the coefficient of -0.027 and p-value = 0.077, stating that firms in group 8 (Utilities) have lower intensity of CSR practices than firms in group 2 (Consumer Staples) at 10% significance level. These results, at a limited extent, are consistent to the findings of Dobre et al. (2015) in case of Romania, and Salehi et al. (2018) in Iran. Briefly, the results for the hypotheses are described in Table 5.

Conclusion
Along with the evolution of global business environment and the public cognitions of social issues, CSR has emerged as the modern corporate strategy for sustainability. However, this concept is multidimensional, vague, and flexible regarding to various interpretations across distinctive contexts, subsequently, generating the enduring debates on how it correlates to corporate financial performance, as well as whether firms should implement CSR activities. Despite a long record of relevant studies, this issue remains fragmented and controversial in emerging countries. Therefore, the present study investigates the relation between CSR and CFP, considering the effect of individual CSR practice to major stakeholders, and the distinctions across industries, using the sample of 162 non-financial listed firms on Ho Chi Minh Stock Exchanges (HSX) during the period between 2015 and 2019, and regression analysis methods. Complimentary tests have shown that the random effects model is the better option for this study.
The results demonstrate that the influence of CSR activities on CFP is inconstant in the context of Vietnam. Specifically, a firm's response to the interest of shareholders and customers can raise its profitability, while its responsibility towards its employees and creditors may lessen its returns. However, the analysis finds no statistically significant correlation between firms' actions towards regulators and suppliers and their accountingbased indicator of financial performance. Moreover, the comparison by industry on the distinction of CSR-CFP relations, though mostly statistically insignificant, finds the lower intensity of CSR practice from firms in Utilities sector compared to firms in Consumer Discretionary and Consumer Staples sectors. Besides, the study finds contradictory influences of different firm-specific factors on CFP. Particularly, while the firm's financial leverage creates a negative impact on its financial performance, the influence of firm size is positive. These findings make great contributions by providing some critical insights for our understanding of CSR practices, supporting firms to establish or adjust their strategic schemes for sustainability in the context of Vietnam.
This study also has certain limitations. Firstly, the study employs regression methods on the panel data of non-financial firms listed on HSX during a five-year period. The consideration of extended time periods, two remaining stock exchanges in Vietnam (HNX and UPCoM), and the use of different analysis methods (e.g., fuzzy Delphi, SEM, or Fuzzy DEMATEL) as suggested by Tran and Nguyen (2021), Tseng et al. (2020), can help to produce more robust results. Secondly, although the study has examined the relation between different CSR activities and CFP, it only takes into account a single accountingbased indicator of CFP and two firm-specific control variables. However, there are potential aspects that may mediate this relation. Thus, it is essential to diversify measures including market-based or survey-based/ qualitative indicators of CFP, and extend independent variables, as suggested by Tiwari and Ahamed (2018), Oktarina and Mu'alim (2017), Darsono et al. (2022). Lastly, many recent theories and studies have suggested a concurrently causal relation between CSR and CFP (Aras et al., 2010;Orlitzky et al., 2003). However, this study is not concerned about potential causation, thus, is regarded as the empirical evidence, rather than causation investigation. Further study can focus on this aspect to contribute the verification of the causal relationship.