Corporate social responsibility and the non-linear effect on audit opinion for energy firms in Vietnam

Abstract Corporate social responsibility (CSR) has increasingly been researched in the current and future trend of sustainable development. Extant studies document mixed results about the relationship between CSR and earnings management, but very few papers discuss how CSR and audit opinion are related. Using a sample of all the energy firms listed in Vietnam from 2007 to 2017, the research results suggest that CSR-engaged firms are more likely to receive unqualified opinions on the quality of their financial statements. We further document that firms benefit from CSR activities only when the activities are not at high levels, or there exists a non-linear effect of CSR on the likelihood of receiving unqualified opinion. This suggests that firms should balance the costs and benefits of CSR. The study conducts several robustness tests and the results remain unchanged throughout these tests.


PUBLIC INTEREST STATEMENT
The matter of social responsibility is no longer the sole burden of public sector, and currently enterprises could feel the need to incorporate socially responsible activities into their strategies. Nonetheless, corporate social responsibility (CSR) could be for window-dressing and decrease firm value, triggering auditors to issue unfavorable opinion on firms' financial statements. Using a sample of all listed energy firms in Vietnam from 2007 to 2017, the current study seeks to examine the link between CSR and audit opinion, and finds that CSR tends to lower the risk of receiving qualified opinion from the auditor. Furthermore, the impact of CSR on the auditor's opinion is conditioned on the levels of CSR engagement: high levels of CSR activities tend to be value destroying, while low levels tend to be value enhancing. The findings in this study imply that stakeholders should be vigilant about the managerial incentives to conduct CSR activities.

Introduction
Firms face so much pressure from various stakeholders that they consider corporate social responsibility (CSR) as an integral part of their actions and operations. In general, CSR could be defined as voluntary activities conducted to achieve social goals (McWilliams & Siegel, 2001). Sustainable development is now a greater shared responsibility and has gaining ground recently (Galaz et al., 2018). Besides ethical motives, previous studies have revealed various CSR motivations, such as risk management (Fombrun et al., 2000) and avoidance of heavy governmental punishment (Spicer, 1978;Cudjoe et al., 2019).
Multiple studies have explored the relationship between earnings management and CSR, because CSR may discourage firms to engage in earnings manipulation. Managers may showcase their integrity and corporate activities could become more transparent when the firm decides to conduct socially beneficial activities. Atkins (2006) reiterates that social responsibility also entails the responsibility to create transparent financial reports. On the other hand, CSR practices can be an instrument for managers to pursue own interest (Jensen & Meckling, 1976). CSR activities may help disguise the consequences of managerial opportunistic behavior (Hemingway & Maclagan, 2004;Jo & Harjoto, 2011;Muttakin et al., 2015;Scholtens & Kang, 2013). In other words, CSR could be positively associated with earnings management.
Since CSR has been shown to have connection with earnings manipulation, it is expected that corporate reporting quality could be affected by its CSR engagement. As a consequence, this may draw the attention of auditors. In contrast with the strong research interest in earnings management, little has been done in the examination of the relationship between firm CSR and audit opinion. This is surprising since audit qualified opinion could result in adverse effect on firm's ability to tap into capital market. If auditors show hesitation and reservation in guaranteeing the financial reporting quality through the issuance of non-standard opinion, lenders will change the way they assess the financial worthiness of the firms, not relying much on accounting-based information (P. F. Chen et al., 2016Chen et al., , 2016. As opposed to western counterparts, developing countries are subject to the impact of severe information asymmetry and agency cost (Stein & Rosefielde, 2005). This is the repercussion of several reasons, some of which are the legal system (La Porta et al., 1998) or weaker corporate governance (Welford, 2007). In environments with severe information asymmetry, auditors tend to play a more important role. Auditors really act as a mechanism to protect investors' interest, as they are able to uncover managerial manipulation through adjusting both income-increasing and income-decreasing accruals (Hirst, 1994).
With regard to the relationship between CSR and audit opinion, L. Chen et al. (2012) show that firms with good CSR performance and low CSR concerns have lower future litigation risk, not only for the firms but also for its auditor, and that auditors tend to charge lower fees and are less prone to issue going-concern opinions to firms with superior CSR performance. However, this study only examines the link between CSR and audit going-concern opinions in a developed country, while the literature remains relatively silent on this relationship in developing countries. In addition, this research lacks the investigation of whether high levels of CSR activities could play a destructive role towards firm performance and lead to firms receiving qualified audit opinions.
In the current study, we seek to extend the current literature by examining the link between firm's CSR activities and the likelihood of firms receiving qualified opinions from their auditors, using a sample of listed energy firms in Vietnam from 2007 to 2017. The sample comprises listed energy firms since these firms are highly sought after for their operational impact on environment as well as sustainable development, i.e., in terms of clean and stable energy source with less waste.
We resort to a measure of CSR by Jizi et al. (2014) and Gray et al. (1995), since this measure takes into account quite a comprehensive set of socially responsible activities. We find that CSR firms appear to be favorably awarded by auditors, i.e. they tend to receive unqualified opinion on the quality of financial statements. Further, considering the non-linear relation between CSR and audit opinion, the study finds that firms benefit from CSR engagement only when the activities are not at high levels, or there is a non-linear effect of CSR activities on the likelihood of receiving unqualified opinion. The study conducts a battery of robustness tests and the results remain robust throughout these tests.
Our study adds to the literature in many ways. First, we extend L. Chen et al. (2012) and provide the empirical evidence on the link between CSR and the likelihood to obtain unqualified opinion from auditors in the context of a developing country. Auditor opinion might play a more significant role for firms in developing countries, so studies that examine the link between these two factors are essential. Second, rather than only discussing the linear effect, we put forward a hypothesis on the non-linear link between CSR and audit opinion, and find robust empirical evidence to support this hypothesis. This result clearly indicates that proper levels of CSR lower the firm's risk exposure and communicate the ethical motives of serving the society through providing transparent financial statements. These lead to firms having lower probability to receive audit "qualified" opinion. On the other hand, high levels of CSR activities may symbolize the tendency of managers to pursue their own interests at the expense of stakeholders, thus raising the likelihood to resort to earnings manipulation to cover up the shirking behavior. This may explain the empirical result that at high levels of CSR activities, firms are more likely to receive qualified opinion, and together it points to a non-linear pattern of the impact of CSR on the likelihood to receive "qualified" opinion from auditors.
The study proceeds with the brief discussion of Vietnamese background on the research objects, literature review and hypothesis development, research methodology, and discussion of research findings. Section 5 will conclude the paper.

CSR and audit opinion in Vietnam
The concept of "Corporate Social Responsibility" (CSR) has been spread into Vietnam through the activities of multinational companies operating in Vietnam. Since then, a number of enterprises have actively implemented CSR, and as a result, their brands have become more visible, such as ACB, Sacombank, Kinh Do, etc. Since 2005, CSR has turned into a noble achievement, and has been honored by trade organizations, such as Vietnam Chamber of Commerce and Industry (VCCI), and Ministry of Labor-Invalids and Social Affairs.
Initially, garment and footwear firms were more interested in attaining the CSR champion glory for its ability to reflect corporate responsibility towards sustainability. The Institute of Labor Science and Social Affairs conducted a survey on firms in the textile and footwear industries, and found that firms in the research sample saw their revenue increase by 25%, labor productivity increase by approx. 10%, and the proportion of export goods also had a marginal rise. Furthermore, the survey suggested that in addition to economic efficiency, firms also benefited from better images with customers, higher engagement and satisfaction of employees, and better ability to attract highly skilled workers.
However, it should be noted that the concept of CSR is still relatively new in Vietnam, and its implementation is still limited. There are barriers and challenges to CSR practices, including limited awareness of CSR concept, lack of financial and technical resources to implement CSR standards (especially for small and medium firms) and confusion due to differences between CSR regulations and the Labor Code, etc. Therefore, CSR is not necessarily associated with higher performance. Nguyen (2018) conducts an empirical examination of the link between CSR disclosure and financial performance of banks in Vietnam in the period of 2011-2016, and concludes that there exists a significantly negative relationship between CSR disclosure and bank performance in Vietnam. The author suggests that this might be the consequences of the burden of the law on social responsibility requirements at the time when banks are in harsh condition as a result of economic downturn.
According to auditing standard No. 700, 1 based on the audit results, the auditor presents one of the following types of opinions on financial statements: unqualified; qualified; refuse to give comments on the financial statements; and refuse to accept the financial statements. Most businesses naturally want to receive unqualified opinion on their financial statements, so that they can receive benefits from the audit report. For example, positive audit opinions can help expand business production by facilitating access to more favorable bank loans. The remaining comments will not be considered as "positive" audit opinion, which will certainly affect the image and reputation of the company. Hoang et al. (2019) examine the effect of earnings quality on corporate social disclosure using a sample of listed firms in Vietnam. The study finds evidence suggesting that there appears to be a positive association between earnings quality and corporate social disclosure, thus rejecting the hypothesis that CSR is related to managerial opportunism. Furthermore, state ownership undermines the relationship between earnings quality and corporate social disclosure, thus providing more justifications for the transition from state ownership to private one.
As Vietnam is still a developing country with fledgling capital markets and inadequate legal framework, the information asymmetry and agency cost tend to permeate throughout the corporate sector, hammering firms' ability to lever on the capital markets. Auditing service plays an important role to ascertain the readability, timeliness, and appropriateness of the financial statements, and unqualified opinion from auditors is crucial to the firm's reputation and prestige. CSR firms tend to outdo non-CSR counterparts and are usually regarded as ethical firms, but whether CSR firms are also responsible in preparing proper financial statements is an empirical research matter.

Theoretical framework
Stakeholder theory is about the strategies to manage groups and individuals that can affect and/or be affected by the focal organization (Mishra & Modi, 2013). Firms are obliged to consider the interests of a wide range of stakeholders in their policies and strategies (Buhr, 2001;Grougiou et al., 2014). The intricate interactions amongst stakeholders make it more ambiguous to understand the firm's operations, exacerbating the information asymmetry issue. Sun et al. (2010) and Grougiou et al. (2014) opine that firms with CSR activities conducted to comfort multiple stakeholders can practice earnings management as a result of high information asymmetry environment. However, another prediction under stakeholder theory is that risk exposure subsides as firms practice CSR (El Ghoul et al., 2011;Godfrey et al., 2009). In satisfying diverse stakeholders' demand, CSR may help reduce the risk of losing one or more stakeholders' supports. This implies lower likelihood to engage in earnings management to gain supports.
Following Legitimacy view, CSR engagement is one of the strategies managers take on to showcase the firm's legitimacy (Grougiou et al., 2014). Firms need to be perceived as legitimate in order to earn social approval to uphold their operations (Suchman, 1995). Deegan et al. (2002) argue that society may terminate the contracts and cease the operations of the firms that lose legitimacy. The legitimacy is earned by conducting activities that are desirable and in line with socially acceptable norms, values, and beliefs. CSR can be a means to enhance the firm's image and suppress social pressure, thus maintaining or even increasing the firm's perceptions of legitimacy (Patten, 1992). García-Sánchez and García-Meca (2017) find that firms are acting responsibly with their CSR activities, thus being less likely to engage in earnings management, which is not a socially acceptable norm.
According to Agency theory, managers have incentives to pursue their own interest due to the separation of ownership and management of a firm. Managers tend to favor opportunistic earnings management to increase the information asymmetry to cover up their unethical behavior. Transparency and accountability obtained from more disclosure of CSR can discourage managers to engage in earnings management (Jo & Kim, 2007). Consistently, Shleifer (2004) and Chih et al. (2008) opine that the stimuli to perform earnings manipulation weaken as information channels are available, and CSR helps to make the firm more transparent. Gray (2007)and Alvarez et al. (2008) argue that firms disclose CSR as a signaling instrument if they are confident in the quality of information and want to differentiate themselves from competitors. Additionally, through CSR disclosure, managers send signal that firms are capable of managing social risks. Lower information asymmetry in turn discourages earnings management, so it is expected that there is a negative association between earnings management and CSR under the perspective of signaling theory.
In general, the four theories predict that CSR alleviates information asymmetry and promote corporate transparency and legitimacy. Therefore, we could predict that CSR may alleviate audit risk in a sense that firms with CSR activities tend to be more responsible towards financial statement users and produce less aggressive earnings. As CSR-engaged firms are less likely to manipulate accruals and real operations, and show lower possibility of receiving litigation claims arising from book-tax earning differentials, audit risk from verifying financial statements of these firms is lower (Lisowsky, 2010). As a result, it is anticipated that firms with CSR activities are less likely to receive qualified opinion.

Empirical literature review and hypotheses development
According to Little and Little (2000) and Godfrey et al. (2009), CSR activities also consolidate firms' reputation and legitimacy, which is an important intangible asset that can produce sustainable tangible benefits. Hong and Kacperczyk (2009) find an unfavorable effect of CSR concerns: firms with controversial CSR performance tend to have thinner shareholder bases, limiting risk-sharing opportunities, and the stocks of these firms are more likely to be priced cheaply compared to those that are more committed to CSR activities.
Additionally, firms with healthy and quality relationships with stakeholders could reduce risk of market uncertainty, lessening any disruption, loss or damage to firm operations and lowering the likelihood as well as impact of undesirable events (Ansong, 2017;Kytle & Ruggie, 2005). One particular relevant explanation for environment-related industries is that environmental and/or social CSR efforts increase the ability of firms to manage and alleviate environmental and other risks, including damages to brand and reputation, boycotts, fines from government, etc.
Studies find the negative link between cost of capital and firm engagement in improving environmental, governance and product attributes (El Ghoul et al., 2011;Ok & Kim, 2019;Sharfman & Fernando, 2008). Viviani et al. (2019) find evidence suggesting that CSR weakens the effect of negative returns on volatility. Investors that advocate green policies also require higher expected returns from polluting firms (Heinkel et al., 2001). CSR helps lower firm risk, so corporate social and environmental spending tends to facilitate firm access to capital market (Cupertino et al., 2019) and better bank loan deals (Goss & Roberts, 2011).
Prior literature offers some explanations on how auditors react to firm's CSR performance. One argument is that since CSR and sustainability engagement can improve firms' foresight, thus enabling firms to weather negative outcomes. In analyzing previous studies, Orlitzky et al. (2003) conclude that CSR helps firm develop competencies in benefiting more and suppressing negativities from external changes, including turbulence and crises, and enables managers to develop a forward thinking style. These skills kick in and the inherent risks of inappropriate management of inventory, accounts receivable, and other asset accounts should reduce.
Besides risk-mitigating incentives, managers may use CSR to enhance the firm's reputation and constrain earnings management to reduce the potential damage to its reputation, which is consistent with a negative association between CSR and earnings management. Lisowsky (2010) finds that CSR-engaged firms have lower likelihood to manage earnings and encounter litigation risks from book-tax differentials. Kim et al. (2012) suggest that more socially responsible firms are more prone to restrict both accrual and real earnings management, thus providing higher quality financial reporting. Lanis and Richardson (2012) indicate a negative link between CSR disclosure and tax aggressiveness. Chen et al. (2012) show that firms with good CSR performance and low CSR concerns are associated with lower future litigation risk, not only for the firm but also for its auditor, and that auditors tend to charge lower fees and are less prone to issue going concern qualifications to firms with superior CSR performance.
In summary, a general theme from previous studies is that thanks to proper stakeholder management capability, CSR tends to curtail risk, and firm performance benefits from CSR activities. Auditing CSR-engaged firms should, therefore, be less risky compared to that of firms that are irresponsible toward environment and society. Accordingly, our first hypothesis is established as follows: Hypothesis H1: Auditors of firms that engage more in CSR activities are less likely to issue qualified opinions.
Nonetheless, merely mentioning the benefits of CSR activities could be misleading. Carroll (2000) argues that firms find it challenging to be both profitable and be an active contributor in social, ethical, and environmental fields. A manager may violate his mandate if he decides to spend excessive amounts of corporate resources on CSR, since Friedman (2007) and Jensen and Meckling (1976) emphasize that a manager's main responsibility is profit. Too many goals put a constraint on firm performance and interfere with firm concentration in earning profits.
CSR is not free and in fact prohibitively expensive, so this could lead to firms losing capital and resources, resulting in firms losing competitiveness compared to those that are not engaged in these activities (Barnett & Salomon, 2006). Additionally, strong CSR commitment raises another agency issue, which is over-investment (Barnea & Rubin, 2010). Heavy investments in CSR activities could build up managerial reputation at the expense of other stakeholders. Carroll (1991) also emphasizes that even when managers are not driven by self-serving motives, their truly moral deeds could negatively affect financial performance because of huge investment in CSR (Wu & Shen, 2013).
Other studies also point out that CSR is not always beneficial, and does not necessarily lead to better firm performance (Gangi et al., 2018;Riyadh et al., 2019;Scholtens & Dam, 2007). In short, CSR could be destructive towards firm performance and lead to auditors discovering improper managerial opportunistic behavior or non-optimal decisions. Therefore, we expect that hypothesis H1 only exists in the case of firms with decent amount of involvement in CSR, rather than those with excessive investment in CSR. In other words, firms with lower amount of CSR activities may benefit more from CSR in terms of audit opinion. Therefore, our second hypothesis is as follows: Hypothesis H2: Firms with lower levels of CSR activities are more likely to garner benefit from CSR in terms of audit opinion.

Research design
The dependent variable in this study is Auditopinion, which is a dummy variable receiving the value of 1 if the auditor issues an unqualified opinion, meaning that the financial reports are considered free from material misstatement. On the other hand, when this variable gets the value of 0, the auditor rates firm's financial statements as not reliable or unable to provide comments on the reliability of the financial statements.
We use a logit model for panel data to identify which variables are important towards the probability of a firm receiving a qualified audit opinion (Gaganis & Pasiouras, 2007;Lynda, 2016;Spathis, 2003). Logistic regression is widely used to investigate the association between a binary response variable and a vector of predictor variables. The main objective of logistic regression is to estimate the mean of the dependent variable, given the value of a set of explanatory variables.
What differentiates logistic regression from linear regression is the fact that the former deals with the nature of the dependent binary variable, rather than continuous one (Fitzmaurice & Laird, 2001). We further control for heteroskedasticity problem with robust standard errors to ensure that the estimates are valid and reliable for interpretation.
The research sample comprises all the energy firms listed in Vietnam from 2007 to 2017. The financial data were retrieved from Thomson Reuters Eikon, while CSR and board size information were hand collected from corporate annual reports. After removing missing observations and trimming outliers following "one per cent" rule of thumb, we end up with an unbalanced dataset of 77 energy firms for the above period, totaling 521 observations.
We draw on extant studies on the determinants of audit opinions (Gaganis & Pasiouras, 2007;Lynda, 2016;Spathis, 2003) to select important variables for the present research. To verify the hypotheses H1 and H2, we include CSR variable in our empirical model as follows: Where π is the probability of receiving unqualified opinion from the auditor. For CSR variable, we follow Gray et al. (1995), Haniffa and Cooke (2005), Scholtens (2009) In each category, we manually check the content of annual reports and award 1 if the answer to whether the firm engages in the CSR activity in each question is yes, and 0 otherwise. Finally, we divide the total score by total number of questions to derive the CSR variable. Therefore, the minimum value of this variable is 0 and maximum 1, with higher values indicating higher involvement in CSR activities by the firm.
Control variables include Size, the logarithm of total assets, for large firms could have better accounting systems and internal control, reducing the likelihood of disagreement with auditing opinion (Gaganis & Pasiouras, 2007;Lynda, 2016;Xie et al., 2010), Leverage, the ratio of total debt to total assets, serves as a proxy for capital strength (Gaganis & Pasiouras, 2007;Lynda, 2016;Xie et al., 2010); Profit, return on assets-ROA, serves as profitability and is added since previous studies find that firms receiving qualified opinions are generally less profitable ones (Gaganis & Pasiouras, 2007;Spathis, 2003), Liquid, the ratio of current assets to current liabilities, is added to control for both potential two-side effect of liquidity on opinion (Gaganis & Pasiouras, 2007;Spathis, 2003); Loss (1 if net income is negative and 0 if otherwise) following (Xie et al., 2010) and BoardNum, number of board members, to control for corporate governance mechanism. AuditCode is dummy variable receiving 1 if the audit firm in charge is one of Big4 firms, comprising Deloitte, Ernst & Young, KPMG, and Price Waterhouse Cooper, and 0 otherwise.
Finally, estimating the impact of CSR on audit opinion could suffer from the endogeneity problem emanating from the two-way relationship between these two factors. Therefore, we employ instrumental regression as well as System Generalized Method of Moments to provide robust results. Table 1 provides general statistics of the variables in the study. The average value of Auditopinion is 0.888, indicating that the majority of the opinions are favorably unqualified ones. The average value of CSR is 0.415; therefore, a firm tends to engage in less than half of the prescribed CSR activities. Leverage is 0.507, or about half of the assets are financed with debt. Profit or ROA is about 7.6%, and Liquid is 2.568, or current assets is about 2.6 times current liabilities, which is a rather high ratio. Loss has the average value of 0.035, or 3.5% of the sample observations experience a loss. Finally, the board size is 5.312, and minimum board size for an energy firm is 3 while maximum is 8, and Auditcode is 0.362, indicating that about 36% of the sample firms are examined by Big4 auditors. Table 2 provides the pairwise correlation coefficients for the variables in the study. CSR has a positive correlation with audit opinion, or higher CSR engagement may lead to better audit opinion. Size has a negative coefficient, which may be inconsistent with the expectation that bigger firms have more decent accounting system, which might limit the probability of receiving qualified opinion from the auditor. Leverage and Loss have negative coefficients, while Profit and Liquid have positive ones, indicating that stronger financial conditions enable firms to have better assessment from auditors. BoardNum has positive correlation coefficient, suggesting that better governance may help firms to obtain more favorable auditor opinion. Table 3 presents the logistic regression result for the whole sample. Firms with negative income are prone to receive worse opinion from their auditors, since Loss has a significantly negative coefficient. Other variables are insignificant. CSR has a positive coefficient, suggesting that client firms with higher engagement in CSR tend to have better and more favorable audit opinion, in line with hypothesis H1. Xie et al. (2010) also suggest that firms with high levels of CSR activities tend to have lower audit fees and less likelihood to receive going-concern opinion. Previous studies have confirmed that CSR and sustainability engagement enhances firms' foresight and planning skills, thus significantly reducing inherent risk in inventory, accounts receivable and other asset accounts (Orlitzky et al., 2003).

Empirical results and discussion
Furthermore, it is found that CSR activities are associated with responsibility towards financial statements so that users have more authentic judgments of the firm's financial position. For example, Kim et al. (2012) find that highly responsible firms tend to restrict both accrual and  real earnings management to render quality financial statements that truly reflect the firm's position. L. Chen et al. (2012) and Lanis and Richardson (2012) find evidence suggesting that CSR activities are associated with lower levels of tax manipulation and lower future litigation risk, so auditors may consider less audit risk when examining these firms.
In summary, on average CSR may signal firms that place emphasis on ethics and responsibility, as well as enable firms to lower the risk exposure to external stakeholders and environment. These altogether render firms to have lower incentives to engage in earnings management and incentivize them to present true picture of firm's financial health. As a consequence, auditors could render less unfavorable opinion and even charge lower fee. Table 4 presents logistic regression result for the two samples split based on their CSR value to test hypothesis H2. Firms with CSR values higher than the median value of the whole sample are considered those that have high CSR engagement, while those that have lower-than-median CSR are deemed those that have low CSR engagement. Loss is again negatively significant, while BoardNum is only negatively significant for firms with low CSR engagement and positively significant for those with high CSR activities. AuditCode is negatively significant, indicating that Big4 auditing firms tend to outperform other auditing firms, and are more likely to discover material misstatements from their client firms. Source: Author's calculation from research data. *, **, *** denote significance at 10, 5, and 1%, respectively. Interestingly, the main explanatory variable-CSR-has a significantly positive coefficient for firms with low CSR activities, and a negative coefficient for those with high CSR values. CSR activities are beneficial to some extent, but are not always so. Excessive engagement as an highly active contributor in social, ethical and environmental fields could sway firms from their main objectives which is to earn profits (Carroll, 2000;Wu & Shen, 2013). In addition, resource constraint forces firms to be more considerate towards the social and environmental expenditures, which may not generate concrete financial benefits, at least in the short term.
The above rationales for the negative link between CSR and firm performance do not mention managerial ethical issues. When the managers are indeed self-serving, excessive CSR commitment raises agency conflict, which is over-investment (Barnea & Rubin, 2010). In this case, managers could be more concerned about enhancing their reputation at the expense of other stakeholders.
The result from the present study are in line with the above studies in pointing out that CSR activities are not always beneficial to firm performance, and may not always embody managerial ethics. When such activities are destructive, especially with high amounts of investment, auditors may feel higher audit risk, which urges them to investigate more closely and, as a consequence, to issue qualified opinion instead. On the other hand, firms with lower amount of CSR activities may balance well between the need to satisfy stakeholders while staying focused on their primary target-being profitable. Therefore, when balancing is obtained, both client firms and auditors may feel lower risk, and the latter are less prone to issue qualified opinion. The evidence here shows support for hypothesis H2.

Robustness check
The results above could suffer from bias when CSR and Auditopinion have mutual relationship. Poor audit opinion could force firms to reconsider their amount of CSR investment from the perspective of agency cost theory. In other words, managers could be denounced to abuse firm's resources for personal reputation following audits, making them unable to persist with the current CSR spending scheme. Also, when firms receive unfavorable comments from their auditors, managers may feel the urge to create a hedge against stock price crashes or other excessively negative consequence through enhancing the current CSR programs.
In the light of the potential two-way connection between CSR and audit opinion, we perform instrumental variable estimation. The instrument used is the average value of CSR for the firms in the sample, since this variable may be well correlated with each firm's CSR variable, while audit opinion of each firm is unlikely to impose an impact on the industry's CSR value. We perform this Source: Author's calculation from research data. *, **, *** denote significance at 10, 5, and 1%, respectively robustness test for the three samples as before, full sample and two samples split based on their CSR values as in Table 5.
For the full sample, CSR still has significantly positive coefficient, confirming that CSR in general plays a positive role in obtaining favorable audit opinion. For sub-sample regression results, we find that CSR for firms with low CSR engagement has higher value (14.267), while that of firms with high CSR engagement has lower value of 10.781. Even though now CSR for high-CSR firms is significant, its lower value compared to that of low-CSR firms is still in consistence with hypothesis H2.
We provide another robustness check to the findings in this study by cleaving the sample into two samples based on CSR values and examining the association between CSR and firm performance in each group (Table 6). If CSR has a more positive impact on firm performance for a sample of low CSR engagement as opposed to that of high CSR firms, this would substantiate the argument that high CSR investment tends to manifest the act of managerial opportunism and futile investments, rather than proper one that adds to firm value. We estimate dynamic model with System Generalized Method of Moments to address the endogeneity inherent to the setup of dynamic model. The coefficient of lagged ROA is significant in both samples, consolidating the use of dynamic model in this study. The p-values of AR(2) and Hansen tests are smaller than 10%, indicating that the instruments are valid and the results are therefore reliable.
CSR is significantly positive (0.107) in the sample of low-CSR firms, implying that CSR helps increase firm performance when it is still low. However, when surpassing a certain threshold, increasing CSR does not lead to higher firm performance, as shown in the sample of high-CSR firms. Together, this piece of evidence suggests that firms tend to benefit from CSR when it is not too excessive. Too much investment in CSR could be the telltale that managers take advantage of firm resources to build up his own fame, or basically CSR is an expensive item yet its contribution toward firm performance is unclear.

Conclusions and implications
CSR is the voluntary activities conducted by firms to achieve social goals, and in addition to ethical motives, previous studies have revealed various CSR motivations, such as risk management and avoidance of heavy governmental punishment. CSR, though, could be detrimental to firm operations, and incentivize managers to manipulate earnings. Nonetheless, in contrast with the extensive research interest in earnings management, little has been done in the examination of the link Source: Author's calculation from research data. *, **, *** denote significance at 10, 5, and 1%, respectively between CSR and audit opinion. This is surprising since audit qualified opinion could result in adverse effects on firm's ability to tap into capital market.
Employing logit regression with robust standard error for a sample of listed energy firms in Vietnam from 2007 to 2017, we find that CSR is positively associated with audit opinion, or firms that engage in CSR activities tend to receive unqualified opinion from their auditors. This is important evidence confirming the beauty of CSR as a risk-mitigating tool and transparencyenhancing channel for firms. We further find that this link only exists for firms that have low levels of CSR, in line with the argument that CSR is beneficial when it is not over a certain threshold.
The practical implications from the findings are that by investing in necessary CSR activities, managers could signal their ethics and care towards society, especially the users of financial statements, thus lowering the chance of receiving qualified opinion from the auditor. Nonetheless, too much investment in unnecessary CSR uses up firm resources necessary for important and profitable projects. Furthermore, CSR could be abused by opportunistic managers to polish their name. These potential rationales should alert auditors as well as financial statement users in assessing firms' motivation to conduct CSR.
The evidence in the current study extends the literature on the impact of CSR. Previous literature has focused on the relationship between CSR and firm performance, and earnings management, but little has been done on its impact on firms' auditor opinion. The study is the first to examine this link in the context of a developing country. Second, the research further sets light on the nonlinear pattern of the influence of CSR on audit opinion. This evidence also suggests a mechanism to reconcile the mixed findings on the link between CSR and firm performance, risk, and earnings management. The research finds that low and high levels of CSR exert different impacts on the probability of auditors providing unqualified opinion on their client firms' financial statements, rather than just a positively linear association between CSR and audit opinion.
In this study, only a sample of listed energy firms in Vietnam in a short period is employed; as a result, the generalizability of the findings suffers. In addition, corporate governance features such as board characteristics or other auditing firm's features including tenure or audit fee could interfere with the impact of CSR on audit opinion. Consequently, future studies could expand the sample to include firms of other industries in both developed and developing markets to verify whether the findings in this study also hold in other settings. Also, future studies could examine how firms of different micro-level and macro-level governance mechanisms, see Widyaningsih et al. (2019) & Iuqman et al. (2018, for example, could benefit from CSR activities.

Funding
This research is funded by University of Economics Ho Chi Minh City, Vietnam.