CORPORATE SOCIAL RESPONSIBILITY AND EARNINGS MANAGEMENT: EVIDENCE FROM VIETNAM

Managers might use corporate social responsibility (CSR) to conceal their misbehaviours. One of these is the implementation of earnings management (EM), which could have serious effects vis-à-vis information disclosed to stakeholders. This study examined the relationship between CSR and EM in Vietnam using the Kinder, Lydenberg, Domini (KLD) scale on five aspects: community, diversity, employees, environment, and products. The results show that in the economic context of Vietnam, for the enterprises in the consumer goods and consumer services industries, those firms with higher CSR scores are more likely to engage in EM. This finding provides additional evidence on the relation between CSR and EM and contributes to streams of studies on earnings and CSR.


INTRODUCTION
Earning is one of the top priorities of businesses.It is not just a goal but also a condition for the existence and development of enterprises.To make accurate decisions, earning is also considered an economic measure for investors to evaluate the efficiency of production and performance of enterprises.As earning is essential, it is not surprising that leaders are always focused on the financial statements of businesses.The financial information of an enterprise is one of the most important sources for sales, production, governance, capital allocation, and capital health information.As a result, businesses might use accounting tools to manipulate financial data to achieve some advantages, known as earnings management (EM) (Healy & Wahlen, 1999).
EM is defined as the use of accounting techniques to produce financial statements that present a positive view of business operations and financial position (Schipper, 1989;Tuovila, 2023).In addition, EM could refer to management practices aimed at positively impacting earnings to avoid reporting insignificant losses and even reporting positive earnings (Dimitropoulos, 2020).EM is conducted for selfinterested reasons, such as receiving compensation, escaping debt agreements, or avoiding regulatory oversights (Choi et al., 2018).Adjustments using accounting techniques could affect financial statements and might impact stakeholders' benefits positively or negatively.For example, enterprises with good financial metrics due to EM might have better values and an improved image, but investors might suffer negative impacts in the long term.In addition, global scandals are also a reminder of the possibility that managers are able to publish unreliable financial information (Ben Amar & Chakroun, 2018).It is worth noting that profit adjustment does not violate legal regulations and is prescribed in accounting standards.
Corporate social responsibility (CSR) is perceived as essential for sustainable growth and better profitability, and CSR activities have been well adopted by enterprises in developed countries.CSR is believed to have an important influence on the relationship between EM and financial performance.Besides, even with EM activities, financial performance can still be improved by CSR (Chakroun et al., 2021).Studies on CSR and EM have received attention in recent years, and their interaction has yielded different results in different countries (Bozzolan et al., 2015;Prior et al., 2008;Rezaee et al., 2020).
In recent years, awareness of CSR in Vietnam has been enhanced with annual CSR Awards presented by the Vietnam Chamber of Commerce and Industry (VCCI).However, its application in developing countries remains a challenge because of the lack of necessary legal frameworks within a weak corporate governance environment, especially in Vietnam, where 97% of businesses are small and medium enterprises (Van Thanh & Podruzsik, 2018); studies on CSR and EM are still limited and have not directly examined the relationship between CSR and EM in the Vietnamese context.Furthermore, previous research results are inconsistent (Prior et al., 2008;Rezaee et al., 2020), leading to inconclusiveness in determining the relationship.
Vietnam is a fast-growing country that had an economic growth rate of 7% in 2019.The fact that Vietnam is a destination for many manufacturing companies after the US-China trade war along with its participation in global free trade agreements (FTAs) has made Vietnam a valuable subject for studies about frontier and emerging markets.Considering the Vietnamese environment, this study will provide empirical evidence on the relationship between CSR and EM in an emerging market.This contributes to the existing literature on CSR and EM with Vietnamese characteristics and expands the empirical literature on the impact of their relationship in emerging economies.Besides, the study is also an additional effort to clarify the importance of a social approach and the need for a set of codes and standards for CSR in a weak corporate governance environment.Moreover, this study could provide meaningful insights on CSR to Vietnamese regulators and stakeholders who are responsible for providing CSR standards and reducing EM.

Theories
CSR and EM are some of the outstanding concepts that have been developed in the past few decades and there are considerable debates over their interrelationship (Chakroun et al., 2021).There are two main approaches-agency theory and stakeholder theory.A conflict of interest between a firm's owners and agents is considered an agency problem.This issue was first mentioned by Jensen and Meckling (1976), stemming from the separation between ownership and control rights.The separation of ownership and management leads to the problem of asymmetric information.Here, the managers (agents) have an advantage over the owner of the information and easily act in their self-interests.Moreover, monitoring the agent's actions is also very expensive, difficult, and complicated.Therefore, managers can hide adverse information for their own benefit by using accounting tools to interfere with the firms' financial statements.In other words, senior managers can take advantage of information asymmetry to perform EM for personal purposes and use CSR as a window-dressing mechanism to conceal misconduct (Choi et al., 2018;Kumala & Siregar, 2020;Prior et al., 2008).This first stream of EM is named opportunistic perspective (Lin et al., 2016).However, in terms of information perspective, the other aspect of agency theory argues that CSR is negatively associated with EM because CSR minimises agency problems in companies by encouraging managers not to manipulate financial statements.In addition, it argues that companies implementing CSR activities tend to be more ethical, honest, and trustworthy.Therefore, firms with high CSR practices are less likely to have EM (Choi et al., 2018;Dimitropoulos, 2020;Kim et al., 2019).Besides, EM might increase agency costs.Other ethical, political, and integrative CSR-related theories indicate that managers in companies implementing CSR are bound to high standards of morals and trustworthiness.As a result, high CSR performing companies are less likely to be involved in the EM activities (Choi et al., 2018;Dimitropoulos, 2020;Kim et al., 2019).
The consequences of implementing EM do not stop at the influence on the owner but also affect stakeholders.According to Freeman (2016), stakeholder theory involves managers of an enterprise being obligated to act in the interests of specific individuals or groups of stakeholders who can influence the organisation.Stakeholders include suppliers, customers, employees, financiers, and the community.Because financial statements do not strictly reflect the financial reality of the enterprise, they can cause wrong decisions by investors and stakeholders.According to Zahra et al. (2005) and DeFond and Jiambalvo (1994), implementing EM in enterprises will deceive stakeholders regarding the actual value of assets, transactions, or financial situation, causing serious consequences for owners, creditors, employees, and society.A fundamental argument of stakeholder-based arguments is that organisations should be governed for the benefit of all members instead of only the interests of shareholders.Stakeholder theory is unified around the motivations of managers to benefit stakeholders by disclosing more information and involving them in CSR activities.Thereby, predicting the relationship between CSR and EM is inverse, meaning that for those enterprises that implement CSR, EM activities might decrease (Dimitropoulos, 2020;Kumala & Siregar, 2020;Prior et al., 2008).

Empirical Evidence
Several studies have shown significant relationships between EM activities and CSR, among which are the studies by Prior et al. (2008)

Hypothesis Development
There are two different arguments for the relationship between CSR and EM according to agency theory and stakeholder theory.Managers on the one hand have the opportunity to achieve personal goals through EM but, on the other hand, due to agency costs as well as managers' motivations to develop relationships with stakeholders, EM activities might decrease (Choi et al., 2018;Dimitropoulos, 2020;Kumala & Siregar, 2020;Prior et al., 2008).From an experimental point of view, research by Rezaee et al. (2020) showed that Chinese enterprises with higher CSR points might be less involved in EM activities than those with low CSR points or no CSR.In addition, another important conclusion is that non-stateowned enterprises are more likely to participate in EM activities.Bozzolan et al. (2015) show that businesses with CSR orientations are less likely to participate in real earnings management (REM) than accrual-based earnings management (AEM).With supporting evidence of the positive effects of CSR on EM, along with the growing interest in CSR in the Vietnamese environment, it is proposed that: H1: CSR has a positive impact on EM.

METHODOLOGY Research Model
To calculate accruals, this study adopted the model developed by Kothari et al. (2005).In particular, Kothari et al. (2005) has shown some limitations of the old model (Jones, 1991) and this is added to the return on assets (ROA) variable to reduce the error of accruals variable to ensure the outcomes of the study.The total accruals are estimated by the following formula: In Equation ( 1), for each i company, TAC i,t is total accruals in year t; TA i,t-1 is total assets in year t-1; ΔREV is receivables in t -receivables in t-1; ΔTR is revenue in year t -revenue in year t-1; ROA i,t-1 is ROA in year t-1.
In this study, EM activities were analysed using the AEM method, which corresponds to economic status and data sample.The research model proposed was based on Rezaee et al. (2020).
In Equation ( 2), AEM i,t is the accrual amount that might be adjusted by management in year t, and the absolute value of the residuals from Equation ( 1) is the proxy for AEM; CSR_SCORE i,t is the CSR rating measured by the total strong score minus the total concern in the five social ranking categories of Kinder, Lydenberg, Domini (KLD) scale; SIZE i,t is the natural logarithm of the firm's total assets; LEV i,t is total debt divided by total assets; MB i,t is market-to-equity ratio, measured as the market value divided by the book value of equity; AGE i,t is the number of years since going public; LOSS i,t is an indicator equal to 1 if the firm incurs a loss and 0 otherwise; MARKET i,t is the natural logarithm of the market index, calculated according to the average VN-Index in the fiscal year.

Variables Dependent variables: AEM
To implement EM using the accrual-based method on the enterprise's financial statements, managers could use the time taken for transactions to happen to record revenue and expenses.Therefore, the data on the financial report will carry a part of the manager's subjective opinion.This proves that implementing the accrualbased method for EM will include two parts: non-adjustable accruals (NDA) and AEM.In particular, the NDA reflects the correct financial value of the enterprise, while the AEM is the value adjusted in the subjective opinion of the managers.Therefore, to identify EM activities by accruals of enterprises, it is necessary to identify accrual accounting variables that could be adjusted by managers (AEM).
In particular, AEM = Total accruals (TAC) -NDA.According to Rezaee et al. (2020), total accruals are calculated with the difference between a business's operating income or net cash flow and business operating activities.The residual value of TAC is used as AEM.

Independent variables: CSR score
The research model's independent variable, CSR_SCORE, was measured by scoring the CSR implementation according to the KLD scale based on five main items: community, diversity, employees, environment, and products (Di Giuli & Kostovetsky, 2014).There were 49 questions to score CSR adopted from the study by Di Giuli and Kostovetsky (2014), with 27 questions about strengths and 22 questions about weaknesses without adjustment.Human rights items were excluded from scoring because of data limitations in the Vietnamese context.The total score was measured by subtracting the total concerns from the total strengths.The authors manually evaluated the final score for each firm using cross-check methodology.The CSR_SCORE rate shows that the enterprise has disclosed information about the implementation of CSR activities based on the items of the KLD scale.Although there are more than 300 internationally recognised guidelines on reporting non-financial information (Gonçalves et al., 2020), KLD was chosen for its compatibility with the information conditions of the Vietnamese market.

Control variables
Based on arguments from previous studies, control variables are included in the model, assuming they have a potential influence on EM activities.Significant differences exist in the conclusions from previous studies regarding the relationship between CSR and EM when the research market is heterogeneous (Prior et al., 2008;Rezaee et al., 2020).Therefore, the market index (MARKET) was examined in the Vietnamese context.Large businesses will be more likely to avoid negative earnings news, and managers might exercise EM to achieve specific goals (DeFond & Jiambalvo, 1994); hence business size (SIZE) should be controlled.Additionally, financial leverage (LEV) was calculated as the difference between total year-end liabilities and the year-end book value of equity, considering debt contracting dynamics to have EM (DeFond & Jiambalvo, 1994).According to the assessment, enterprises that are highly appreciated by the market will receive greater attention from investors, thus limiting the behaviour of EM and vice versa.As a result, the market rate on equity (MB) should be examined (Rezaee et al., 2020).It is argued that new businesses will try to display a good image based on the implementation of EM to attract the attention of investors, and the number of listed years (AGE) is taken into consideration (Rezaee et al., 2020).Finally, business efficiency (LOSS) is an attribute of enterprises that implement EM to create better values in terms of accounting.Still, it is not synonymous with the enterprise's business performance (Rezaee et al., 2020).

Robustness check: Earnings persistence
A robustness check is performed to examine whether estimated regression coefficients are reliable by using earnings persistence as a proxy for earnings quality.Earnings persistence could be considered as superior earnings to ensure a permanent part of the future earnings (Rezaee et al., 2020).The earnings persistence model of Rezaee et al. ( 2020) is followed as in the below equation: In Equation (3), ROA i,t is net income divided by total assets for firm i in fiscal year t, and ROA i, t−1 is net income divided by total assets for firm i in fiscal year t-1.α 1 measures earnings persistence before CSR consideration, and interaction coefficient α 3 measures the impact of the CSR_SCORE on earnings persistence.

Sample
Information was collected from 87 enterprises in consumer goods industry groups published on the Vietnamese Stock Exchange in the three years of 2017, 2018, and 2019.This was the high growth period of Vietnam's economy before a sudden decline in 2020 due to the uncertainty of the pandemic.CSR scores were rated based on a KLD scale forming balanced panel data.This provided observations of the same company in all periods, reducing the noise introduced by unit heterogeneity.
Companies in the industry were chosen because of large fluctuations in receivables.
Those selected are likely to take EM activities to maintain the company's condition in the industry (Triliawati, 2020).
Data was collected based on two primary sources: financial statements published on the business website and data from the FiinPro data system and reputational websites of CafeF and Vietstock.Data from these sources were used to calculate dependent variables and control variables.

Descriptive Statistics
According to the descriptive statistics, the significant differences in the values of the AEM show that the EM implementation is different in businesses.In particular, AEM values range from -0.2 to 0.5, showing two different aspects in EM.If the value of AEM is negative, the enterprise shall perform revenue reduction by shifting a partial turnover to the following year.It could be explained that enterprises aim to reduce corporate income taxes over the year.On the other hand, in case the value of AEM is positive, the enterprise performs a revenue increase by booking revenue before the actual time or shifting cost.These conditions create incorrect financial information that affect investor decisions (Zahra et al., 2005) or cause more preferential conditional loans when banks assess the ability of businesses (DeFond & Jiambalvo, 1994).For the value of CSR_SCORE, a considerable difference exists for businesses that are evaluated according to the KLD scale.Theoretically, the maximum CSR_ SCORE value that can be achieved is 22.However, in the study's selected industry group, the maximum is 15, and the lowest is 0. The average value of 4.65 also shows that most enterprises have a relatively low CSR_SCORE compared to the maximum value of the observed variables.The values of the observed variables show that the implementation and disclosure of CSR-related information of some enterprises in the consumer goods and service industries are mainly at a relatively low level.

Correlative Analysis
Pearson correlation analysis was done to determine the linear correlation relationship of variables with the description (Table 2).Through the Pearson correlation matrix of the research variables, the r coefficient ranges from -0.1475 to 0.4718.The dependent and primary independent variables have coefficient r = 0.1340 (relatively low), so multicollinearity is very unlikely.The CSR_SCORE and SIZE variables have the highest correlation coefficient (r = 0.4718).This is reasonable because large-scale enterprises often have better CSR policies for different purposes, such as communication or community support, than small companies.In many cases, smaller enterprises still have CSR policies but are more limited than larger enterprises in terms of public relations.The SIZE and LEV also have a high correlation, which can be explained because it is easier for large-scale enterprises to access and take advantage of debts.

Multivariate Analysis
Regression analysis was also performed to determine the relationship between CSR and EM in the study.Table 3 describes the results of the cross-sectional regression equation (2) using Fixed Effect and two-stage least square (2SLS).
Considering the potential problems of endogeneity in CSR and EM association, an appropriate approach is required (Choi et al., 2018;Dimitropoulos, 2020;Kim et al., 2019;Palacios-Manzano et al., 2019).The endogeneity problem could be solved using panel data and control fixed individual effects.A 2SLS approach was also employed to have prudent results.Additionally, a lagged CSR variable (one period) was used as an instrument similar to studies of Palacios-Manzano et al.
(2019) and Dimitropoulos (2020).Notes: AEM i,t is accrual amount might be adjusted by management in year t; EP is measuring by ROA i,t which is net income divided by total assets for firm i in fiscal year t; ROA i, t−1 is net income divided by total assets for firm i in fiscal year t-1 CSR_SCORE i,t is CSR rating measured by the total strong score mined with the total concern in the five social ranking categories of KLD ranking data: community, diversity, employees, environment, and products; SIZE i,t is the natural logarithm of the firm's total assets; LEV i,t is total debt divided by total assets; MB i,t is market-to-equity ratio, measured as the market value divided by the book value of equity; AGE i,t is the number of years since public; LOSS i,t is an indicator equal to 1 if the firm incurs a loss and 0 otherwise; and MARKET i,t is a natural logarithm of the market index, calculated according to the average VN-Index in the fiscal year.Data samples (Dimitropoulos, 2020)  Regression results show that CSR_SCORE variables are statistically significant for both Fixed Effect and 2SLS estimators in the AEM model.In addition, the earnings persistence demonstrates similar results where coefficients for interaction terms between earnings and CSR rating are statistically significant for both Fixed Effect and 2SLS.
CSR_SCORE reached statistical significance at 1% for Fixed Effect and 10% for 2SLS levels and the coefficients were positive.This shows that for enterprises in the consumer services and consumer goods industries in Vietnam (for the period 2017-2019), the higher the CSR rating, the higher the EM activity in increasing income or reducing losses, if any.This result is consistent with Prior et al. (2008), who explained that EM implementation should personally benefit managers.This, however, is inconsistent with the results of Rezaee et al. (2020), Dimitropoulos (2020), Kumala and Siregar (2020), Gonçalves et al. (2021), andGaio et al. (2022), where firms with higher CSR scores are less likely to engage in EM activities.This result provides a warning signal about certain activities aimed at enhancing CSR in the Vietnamese environment.
This result contradicts the research hypothesis H1 but is also a new perspective on the relationship between the two research variables.The results support agency theory in terms of opportunistic perspective.A business that shows good CSR can help create an excellent image to stakeholders to cover up its behaviour of EM implementation.This is still considered a relevant and valid result.This does not represent all businesses but may still represent the majority in the consumer goods and services industry.
SIZE reached statistical significance levels at 1% for the Fixed Effect and 2SLS estimates.It can be argued that the larger the firm size (by asset value), the less likely it would engage in AEM activities.Furthermore, larger enterprises could receive more significant interests from stakeholders and, therefore, EM activities could be controlled more than those of small businesses.This result is consistent with the finding of Rezaee et al. (2020) but inconsistent with conclusions made by Kumala and Siregar (2020) and Dimitropoulos (2020).
The MB ratio reached statistical significance at 5% for 2SLS.Therefore, it can be reasoned that firms with a high MB ratio are less likely to engage in AEM, or in other words, firms that are appreciated by the market are less likely to participate in more AEM.This result may be due to the fact that these companies are more likely to be scrutinised by regulators and investors and, as a result, motivate them to maintain high-quality financial statements, and less likely to perform EM.It is similar to the outcome of Dimitropoulos (2020) but inconsistent with the findings of Rezaee et al. (2020).
The MARKET reached statistical significance at 5% for Fixed Effect.It can be argued that when the MARKET increases, the more likely companies would be to engage in AEM activities.It could be argued that companies might want to take advantage of market opportunities to increase resources through EM.This contradicts Rezaee et al. (2020) where the moderating effect of the marketisation index was found.Control variables LEV, AGE, and LOSS had no statistical significance.

DISCUSSION
In the economic context of Vietnam in the period of 2017-2019, businesses in the consumer goods and consumer services industries that are evaluated well for their social responsibility implementation tend to perform accrual-based earnings management.These include all five aspects of community, diversity, employees, environment, and products.As a result, the information provided in the financial statements may mislead stakeholders and increase long-term risks.In addition to examining the relationship between CSR and EM, this study implemented social responsibility scoring through the KLD scale with 49 questions, and is the first study in Vietnam to accomplish this assessment.
In the context of the available literature, this study provides a new perspective for researchers on the issue of implementing CSR and EM activities in Vietnamese enterprises.This study offers an empirical contribution to the existing literature of CSR and EM with findings from an emerging country.Unlike previous studies conducted in other markets where high CSR performing firms are associated with fewer EM activities (Choi et al., 2018;Dimitropoulos, 2020;Kim et al., 2019;Kumala & Siregar, 2020;Palacios-Manzano et al., 2019), this research shows the opposite relationship.This study supports the argument of agency theory that, due to asymmetry information, senior managers might sacrifice stakeholders' interests for their own interests through the implementation of CSR to create a noble image of the company and conceal the EM activities.It should be noted that the institutional context of Vietnam is an important point.Indeed, with the complete absence of a supportive regulatory framework and a weak corporate governance environment, it is unlikely that CSR will play a significant role in controlling manager behaviours and, therefore, affect the restriction of EM.
These results suggest several implications for regulators, accountants, auditors, investors, and researchers in a number of countries.In practice, the study notifies business stakeholders of the possibility that their long-term interests may be affected, thereby increasing the level of precaution before making decisions based solely on the CSR of an enterprise.As a result, users of financial statements must be extra cautious when analysing companies with more CSR activities.The results also provide valuable insights for regulators to improve market transparency by greater incentivisation of CSR.The results encourage regulators to introduce stronger rules and investor protection mechanisms to restrict EM activities.The results also suggest the need to implement effective audit practices, especially the effects of audit quality on AEM.Those are especially useful in the absence of a suitable legal framework for both CSR and EM.

CONCLUSION
This study is a preliminary effort to address the relationship between CSR and EM in the Vietnamese context where firms with high CSR are found to be associated with higher EM activities in the consumer goods and consumer services industries.
Unfortunately, the study has some limitations.Firstly, it only considered businesses in the consumer services and goods industries in 2017-2018, which is not a general pattern for the whole economy.Secondly, even though the KLD scale is one of the popular scales of social responsibility, it is not yet sure of its appropriateness in the economic context of Vietnam.Therefore, more studies using other scales are needed to assess the appropriateness of the scale.In addition, it is necessary to use different methods of measuring EM in addition to the AEM and EP.Lastly, related studies on the topic of CSR and EM are considered a new perspective.Therefore, it is possible to develop the study by expanding the number of samples and control variables that can effectively represent Vietnamese or regional businesses.

Table 1
Research model variable descriptive statistics

Table 3
Regression on the relationship between CSR and EM include 249 observations from 87 firms listed in Vietnamese Stock Exchange for 2017, 2018, and 2019., ** , and *** indicate statistical significance at the 0.10, 0.05, and 0.01 levels, respectively, based on two-tailed tests *