Corporate Social Responsibility and the Misclassification of Income Statement Items during the Coronavirus Pandemic

Abstract

The purpose of this study was to examine the relationship between corporate social responsibility (CSR) and earnings management in Gulf Cooperation Council (GCC)-listed companies. It specifically addresses the question of whether companies that practice greater corporate social responsibility are less likely to engage in earnings management practices. The study sample consisted of 300 firms listed between 2015 and 2021 on GCC bourses (Saudi Arabia, United Arab Emirates, Bahrain, Qatar, Oman, and Kuwait). In this study, we developed multiple linear regression models and collected data from the Bloomberg database, Refinitiv, annual reports, official firms’ websites, and the GCC’s bourse websites for the period from 2015 to 2021. In the pre-pandemic period, firms that engaged in corporate social responsibility activities were more likely to have fewer classification-shifting practices. During the pandemic era, however, this relationship became significantly positive, suggesting that firms’ corporate social responsibility practices may be used to hide their opportunistic classification-shifting practices during difficult times, such as a pandemic. This paper presents a thorough investigation of how businesses may alter their behavior toward increasingly applied but understudied earnings management strategies and CSR practices during a difficult period such as a pandemic.

1. Introduction

The worst physical, social, and economic calamity in history is said to have been caused by the coronavirus (COVID-19) pandemic (Lassoued and Khanchel 2021). In an effort to stop the COVID-19 virus from spreading globally, various nations implemented lockdowns and severe controls; nonetheless, this slowed down the international economy and negatively influenced the operations of several businesses (Barai and Dhar 2021). The pandemic was an unprecedented crisis that led to an economic downturn (Jordan et al. 2021). Previous studies documented that managers are more likely to manage firms’ earnings during financial distress (Trombetta and Imperatore 2014; Jordan et al. 2021). Healy and Wahlen (1999) defined earnings management (EM) as “when managers use judgment in financial reporting and in structuring transactions to alter financial reports to either mislead some stakeholders about the underlying economic performance of the company, or to influence contractual outcomes that depend on reported accounting numbers”. Given the COVID-19 scenario and how companies may behave during financial crises, it was anticipated that firms would engage more in EM practices in order to adapt to the underlying unfavorable market conditions (Jordan et al. 2021). Although EM would provide better financial conditions for firms during a crisis, it could result in severe economic consequences in the long run such as increased operational risks within businesses, weakening their financial stability, and escalating the cost of capital (Kim et al. 2021). Therefore, the previous literature perceived EM as an unfavorable practice (Zalata et al. 2019) and viewed it as unethical because it distorts the true financial information of firms and provides a fake picture of firms’ financial reports to investors (Belgasem-Hussain and Hussaien 2020). Furthermore, Hamilton et al. (2018) surveyed 122 public company managers with financial reporting experience and the results revealed that managers consider EM to be relatively immoral and culturally unacceptable. In addition, Velte (2020) emphasized that there is a slight difference between EM and financial fraud, as financial fraud appears as an extreme form of EM.

On the other hand, CSR has drawn much attention and grown to be one of the most hotly debated subjects in academia as companies become more conscious of their moral and ethical behavior (Kolsi and Attayah 2018). Since EM practices are expected to increase during a pandemic (Xiao and Xi 2021), the research has indicated that there is a sharp increase in the need for CSR initiatives (Manuel and Herron 2020). This is because financial transparency and accountability are fundamental principles of CSR, and corporations are required to participate in CSR initiatives to provide reliable financial data to different stakeholders (Kim et al. 2012; Mohmed et al. 2020). Prior researchers concurred that companies with a high level of commitment to CSR are expected to be less likely to alter their earnings, as they tend to be more accountable in their preparation of financial statements (Kim et al. 2012; Choi et al. 2013).

Even though stakeholders have given CSR much attention, previous research has focused more on the relationship between CSR and corporate performance than on the relationship between CSR and EM (e.g., Nollet et al. 2016; Alsartawi 2020). It is crucial to look at the connection between CSR and EM activities, as EM practices have the potential to cause value distortion just like what happened with the Enron and WorldCom scandals (Dechow et al. 1995; Karpoff et al. 2008; Chen et al. 2015). Furthermore, the results of earlier research linking EM and CSR are still unclear. For instance, some studies found a negative link (e.g., Cho and Chun 2016; Alsaadi et al. 2017), while other studies’ findings indicated a favorable relationship (e.g., Scholtens and Kang 2012; Muttakin et al. 2015; Kolsi and Attayah 2018; Garfatta 2021; Mohmed et al. 2020).

However, our research fills the gap in the existing literature on conflicting ideas for the following reasons: The EM literature has classified EM practices into three types: the accruals earnings management method (AM), real earnings management (RM), and the misclassification of income statement items (CS). Previous research has focused much attention on a particular type of EM, namely the AM (e.g., Al-Maliki and Abdul Rahman 2019). This method focuses on managing firms’ accruals, and it has attracted the attention of regulators and policymakers. The other EM methods, such as CS, however, have not received as much attention. As managers employ a range of EM strategies in addition to the AM, this only presents a partial representation of the actual financial condition of a company (Abernathy et al. 2014; Zhu et al. 2015). Researchers have found that while the usage of the AM has been decreasing over the past few decades, the use of different additional EM methods has increased dramatically, especially following the global financial crisis in the year 2008 (Anagnostopoulou and Tsekrekos 2016). Additionally, this study makes the case that all EM practices, regardless of risk level, should be curtailed if businesses are to be considered socially responsible. However, if CSR activities are positively associated with a complex and a less risky EM method (i.e., CS), then firms are not ethically sensitive, as all EM practices are deemed unethical (Zalata and Roberts 2016).

In addition, if previous research indicates that the AM and CSR have a negative relationship (Almahrog et al. 2018; Gaio et al. 2022) and do not include other EM techniques, this will result in providing missing information about other EM practices and, accordingly, contribute to firms’ value distortion. To learn more about the relationship between EM and CSR, it is important to shed light on the various EM techniques such as CS, as the literature documents that this particular practice has increased tremendously in recent years (Bansal et al. 2021). Furthermore, a review of past studies showed that most of them focused on developed countries, such as the United States and the UK (e.g., Grougiou et al. 2014; Almahrog et al. 2018; Dimitropoulos 2022), while a lesser amount of attention was paid to underdeveloped countries. For example, only a few studies have looked at this association in the Gulf Cooperation Council (GCC) region (e.g., Gerged et al. 2018; Kolsi and Attayah 2018; Garfatta 2021). As stressed by Menassa (2010), the GCC is one of the leading Arab economies and it would be intriguing to test this association in this region. Moreover, Kolsi and Attayah (2018) claimed that earlier studies that looked at this relationship lacked comprehensive evidence because emerging countries were unaware of the importance of CSR activities, and there was no clear CSR guideline.

However, the GCC countries have implemented diverse measures to enhance and maintain CSR practices. To enhance the intellectual capital of firms, for instance, a number of scholarships were offered to workers, students were supported, an environment conducive to entrepreneurship was fostered, awards related to CSR objectives were offered, and firms were heavily involved in charitable and donation activities (Aldosari and Atkins 2015). In addition, regulators in this region are considering the benefits of CSR engagement more frequently. Given that the findings from empirical studies are still tentative, more investigation is required to ascertain how CSR efforts in the GCC can affect EM practices.

This study aims to close this gap by highlighting the relationship between CSR and EM in GCC-listed firms. This study offers a more comprehensive understanding of the connection between CSR and EM by focusing on a particular method that received less attention in previous research, which is the misclassification of income statement items to inflate core earnings (CS). Studying the CS method in the GCC is essential as this region follows a principle-based approach (i.e., international financial reporting standards) (IFRS). Under IFRS, constraints concerning non-recurring items in income statements are not as strict as under GAAP (Zalata and Roberts 2016); hence, managers have more freedom to choose how to classify both revenue and expenses in income statements to their benefit. Therefore, managers are more likely to utilize this method than the other EM methods (i.e., the AM and RM). Second, the GCC countries are known for their underdeveloped governance environment and require comprehensive scrutiny (Jizi et al. 2022); hence, the chance for managing CS is probably higher. Third, since core earnings reporting has gained popularity as an indicator of performance in financial markets, lawmakers are paying greater attention to it (Rapoport 2016). Especially after the recent decline in oil prices, policymakers in the GCC are aggressively exploring ways to gain investors’ trust and reduce their reliance on oil. Therefore, focusing on the CS method is important to reveal the level of reporting reliability.

In addition, despite being less risky, the CS technique may severely harm investors by misleading them about the future profits of companies (Anagnostopoulou et al. 2021). Additionally, like the AM, deliberately managing core earnings can have an adverse effect on the future performance of operations and cash flows (Cain et al. 2019). Furthermore, as stressed earlier, prior studies focused on linking CSR to the AM and RM, while the CS research area is still understudied although it was documented that the attitude toward misclassifying income statements has recently increased (Anagnostopoulou et al. 2021). Furthermore, previous research studies concurred that endogeneity issues might surface in corporate governance research (Velte 2020; Zaman et al. 2022); hence, this work employs many robustness tests to resolve the variability issue.

Moreover, this study provides a unique argument regarding ethics versus values by looking at CSR and EM practices from different angles. This study suggests that socially responsible companies are more likely to engage in CSR activities and less likely to engage in CS, as, although it is a less risky EM strategy than the other EM strategies, it is still an unethical practice. This assumption supports the stakeholders’ theory argument, as engaging in these unethical behaviors will mislead the interested parties. However, if businesses participate in both CSR and CS, the results will be consistent with the agency theory’s viewpoint on managerial opportunistic behavior, as managers would be participating in CSR practices to hide their opportunistic CS practices. Additionally, by determining whether the COVID-19 pandemic affected enterprises’ choice of CS, this study aims to expand upon previous EM research by examining whether managerial behavior differs during an exceptional and difficult time.

The remainder of this article is organized as follows. Section 2 reviews the relevant literature and presents our research hypothesis. Section 3 describes our data, sample selection criteria, and research methods. Section 4 presents the main empirical results and additional analysis results. Section 5 concludes this study.

2. Literature Review and Hypotheses Development

Corporate Social Responsibility and Earnings Management

Many researchers and international bodies have tried to provide a precise definition of CSR. For instance, Carroll (1979) defined CSR as “businesses’ obligations to society’s legal, moral, ethical, and philanthropic standards”. According to the World Business Council for Sustainable Development, a company’s commitment to sustainable development is demonstrated by its CSR disclosure. The United Nations Industrial Development Organization defines CSR as “a management concept in which companies integrate social and environmental concerns in their business operations and interactions with their stakeholders” (UNIDO). A more contemporary definition of CSR was provided by Huang and Watson (2015), who said that it was related to companies carrying out socially conscious operations in addition to legally required ones. It can be noted that some definitions focus on the social side of CSR, while others include other aspects such as the environment. There is a substantial connection between the activities required to achieve the Sustainable Development Goals (SDGs) and the thematic development fields of CSR. The SDGs are shown in Figure 1. CSR has a comprehensive framework and provides guidance for achieving a more sustainable future, while the SDGs establish specific and clearly stated goals to assess the results of efforts.

According to Kolsi and Attayah (2018), as CSR is associated with the moral and ethical aspects of business conduct and decision-making, it would both reduce financial risk and raise stakeholder satisfaction. However, very few studies (e.g., Cao et al. 2016; Mohmed et al. 2020) have looked at the relationship between CSR disclosures and EM; much earlier research on CSR focused on characterizing and indexing CSR as well as evaluating its effect on business performance. The relationship between EM and socially responsible businesses must be studied, though, as EM hides and distorts a company’s true financial performance (Fan et al. 2019) and is viewed as a paradigm for evaluating corporate ethical issues (Heinz et al. 2013). Earnings are a crucial factor in valuing businesses, so EM has become a constant issue for practitioners and policymakers in addition to academics (Dichev et al. 2013; Walker 2013; Kourdoumpalou 2017; Sanad et al. 2022).

Many studies have attempted to provide a broad definition of EM; Grimaldi et al. (2020) noted that, considering the intricacy of EM, this can be difficult, and they claimed that EM is a “grey rule” that may provide “beautiful” financial results while adhering to standard accounting procedures (Zhou et al. 2020). The three types of EM approaches that have been covered in the literature are the AM, RM, and CS, according to Walker (2013) and Malikov et al. (2018). Studies agree that any method used to influence a company’s results might exacerbate information asymmetry between management and interested parties, hide the real state of the company’s finances, and undermine the validity of financial reporting (Zalata and Roberts 2016; Eiler et al. 2021).

The timing, risks, and costs involved are just a few of the factors that affect managers’ preference for EM methods (Anagnostopoulou et al. 2021). Contrary to the AM and RM, CS has not been extensively studied in the literature for many years, but a growing number of EM research studies have begun to do so. Malikov et al. (2018) asserted that CS is a “recent form of earnings management”. Because it has no impact on a company’s bottom line earnings since it focuses on modifying the firms’ core earnings only, CS may not seem to be as harmful as other EM types, but it has grown to be a major EM concern. According to Fan et al. (2010) and McVay (2006), CS would eventually result in the deception of stakeholders, especially if managers are unwilling to employ other EM techniques.

According to Malikov et al. (2018) and McVay (2006), CS is the misclassification of income statement items that do not affect profits at the bottom of the business. Skousen et al. (2019) defines CS as the purposeful reporting of income statement items across many lines. A more comprehensive definition of CS was provided by Poonawala and Nagar (2019), who said that it is the deliberate misrepresentation of revenues or costs on the income statement in order to increase a business’s gross profit or primary profit while maintaining net income. Alfonso et al. (2015) said that CS is a creative and useful approach to controlling profits.

For CS, a range of methods can be applied. As proposed by McVay (2006), managers may purposefully shift operational expenses (such as the cost of products sold and sales, general, and administration expenses) to special items in order to increase core profitability. In an attempt to increase core earnings, managers erroneously move operational expenditures to the ceased operations area, which reduces income (Skousen et al. 2019). Since IFRS is founded on principles, managers are free to categorize items on the income statement to let readers of financial reports know whether they are one-time or ongoing items, as well as to let them know that a firm’s earnings are predictable and consistent, which is a sign of high earnings quality (Orjinta 2018). Furthermore, because IFRS rules surrounding non-recurring items in the income statement tend to be less strict, managers are afforded greater flexibility when it comes to classifying costs and revenues within the income statement (Zalata and Roberts 2016; Chung and Chae 2020). As per Gray et al. (2015), managers are permitted to exercise their discretion to manage a corporation’s revenues to their benefit under principle-based criteria, provided that they utilize their judgment to attain the optimal economic situation of the organization.

CS differs from the other two EM techniques in a number of ways, but they all aim to mislead those who are interested in the financial condition of firms (Zalata et al. 2019). All three of the EM strategies that managers employ—the AM, RM, and CS—have the ability to mislead investors about how well a firm will do in the future (Anagnostopoulou et al. 2021). Just like with the AM and RM, purposeful core earnings management might negatively impact operating performance in the future (Cain et al. 2019). As a result, regulators have lately given core earnings reporting their full attention as it has grown to be a well-liked performance metric in capital markets (Rapoport 2016).

Two lines of prior research may be distinguished in terms of the connection between CSR and the EM controversy (Prior et al. 2008; Kim et al. 2012; Mohmed et al. 2020). Some research has explained the link by pointing to stakeholder and stewardship theories and asserting that engaging in CSR activities would enhance the quality of profits. Choi et al. (2013) suggested that companies with a strong CSR profile are probably more accountable for their financial statement preparation. Moreover, several scholars agreed that companies that participate in CSR initiatives exhibit authentic consideration for the concerns of diverse stakeholders, leading to enhanced profit quality (Cao et al. 2016).

According to agency theory (Jensen and Meckling 1976; Fama and Jensen 1983), companies may participate in CSR initiatives to improve their reputation with stakeholders, but they would also be pursuing their interests (Choi et al. 2013), such as hiding their opportunistic behavior or misbehavior (Mohmed et al. 2020). In other words, businesses that participate in CSR programs are more likely to generate low-quality profits. Testing the link between CSR and EM is difficult as a result of the controversy (Prior et al. 2008; Mohmed et al. 2020). Furthermore, there are still contradictory scientific findings in this area. For instance, EM and CSR have been reported to correlate negatively in a number of studies (e.g., Kim et al. 2012; Choi et al. 2013; Cho and Chun 2016; Alsaadi et al. 2017; García-Sánchez and García-Meca 2017). However, a few studies (e.g., Gargouri et al. 2010; Scholtens and Kang 2012; Mohmed et al. 2020) have discovered a favorable correlation.

Due to the ambiguity surrounding regulations and the function of CSR in businesses, there is a dearth of studies conducted in the GCC analyzing the connection between CSR and EM practices (Kolsi and Attayah 2018). However, GCC businesses and communities are becoming more cognizant of CSR (Al Ani and Jamil 2015; Alsartawi 2020; Ghardallou and Alessa 2022). According to Shehata (2015) and Alhussein and Alenaze (2024), the GCC countries are starting to realize the importance of corporate social responsibility. For example, Kolsi and Attayah (2018) used an Abu Dhabi Stock Exchange (ADX) sample to investigate the relationship between CSR and two EM approaches, the AM and RM, in the UAE setting. The study discovered a favorable link between the AM and CSR disclosure levels but no correlation between CSR disclosures and RM behaviors. Moreover, Garfatta (2021) investigated the link between EM and CSR disclosure in Saudi Arabia and discovered a strong positive correlation between the two. A negative correlation between CSR and the AM was discovered by Gerged et al. (2018) using a sample of Kuwaiti stock market -listed businesses. Furthermore, Al Ani and Jamil (2015) discovered that CSR disclosure only affects the AM in Kuwait and Bahrain, based on a sample of nonfinancial listed businesses from the GCC markets. Also, Aljifri and Elrazaz (2024)’s study findings revealed that employing the AM method has a detrimental effect on the quality of earnings for both financially distressed and financially stable companies in the GCC region. However, it has a beneficial effect on the long-term viability of earnings for all types of firms.

Prior research examining the relationship between CSR and EM in the GCC is limited and the findings are mixed. Moreover, most of the studies employed the AM, a particular type of EM, and very few considered alternative EM methodologies. Furthermore, the bulk of research concentrated on a single GCC nation, with a few exceptions (e.g., Al Ani and Jamil 2015). Because of this, this study adds to the body of evidence explaining the relationship between CSR and CS as a particular form of EM that has not received much attention in the literature, even though, as already mentioned, CS practices have grown recently, and their primary purpose is similar to that of the AM method—that is, to misinform shareholders by altering firms’ earnings to achieve opportunistic managerial goals (Zalata et al. 2019). The existing literature concurs that the CS method is linked to low litigation risks due to its association with high managerial discretion, thereby limiting the auditors’ capability to validate it (Alfonso et al. 2015; Zalata and Roberts 2017). Additionally, since core earnings reporting has become a more preferred metric by shareholders in capital markets, regulators are now paying more attention to it (Rapoport 2016). Despite being referred to as a “soft EM method”, the CS method could result in serious repercussions because it could misinform investors about how a business will perform in the future (Anagnostopoulou et al. 2021). Additionally, like AM, handling core earnings deliberately could have a detrimental effect on subsequent years’ profitability and cashflows (Cain et al. 2019). Since this issue has not been addressed before in the GCC context, it is important to look at EM from a variety of angles.

CS is less risky than other EM practices; hence, this study suggests that if firms are truly ethical and socially responsible, they will likely engage in CSR activities while being less likely to engage in CS practices. This supports the stakeholders’ theory argument because engaging in these practices would mislead the interested parties regardless of the risk level of the EM method. However, the results would be consistent with the agency theory’s opportunistic behavior perspective if businesses participated in both CSR and CS activities. In light of the conflicting results of other investigations, the current study upholds the following non-directional hypothesis:

H1. 

There is a significant correlation between EM and CSR.

Additionally, the literature has discussed the economic effects of the COVID-19 pandemic, which was one of the most significant crises to have occurred and led to lockdowns and social isolation, thus decreasing revenue for many businesses. Businesses faced difficulties in surviving and improving their financial performance during the pandemic (Carracedo et al. 2021). As a result, businesses are more likely to use EM practices during a crisis or pandemic (Ozili 2020), and there is a stronger need for CSR initiatives in response to the pandemic (Manuel and Herron 2020).

In line with El-Feel et al. (2024), this study argues that companies tend to manipulate their earnings during a crisis by intentionally reporting higher earnings, as observed in the year affected by the pandemic.

Because investors rely more on core profits to make investment decisions, this study suggests that businesses are more likely to participate in CS activities in an effort to inflate their core earnings and provide a better picture of their financial performance (Chen et al. 2015). Additionally, as mentioned earlier, the stakeholders’ theory supports the notion that firms would engage more in CSR activities and less in EM practices (Alsaadi et al. 2017). Similar to our previous argument, CS is a comparatively less risky approach than other EM practices. This study indicates that companies that are genuinely ethical and socially responsible are more inclined to participate in CSR activities, while being less inclined to engage in CS practices, especially during a difficult time such as a pandemic. Nevertheless, if businesses engaged in both CSR and CS activities during a crisis, the outcomes would align with the opportunistic behavior perspective of agency theory (Wasiuzzaman et al. 2015). This study implies that companies could have increased their CSR activities during the COVID-19 period to compensate for their deceptive actions. Hence, the second hypothesis is formulated as follows:

H2. 

CSR and EM had a substantial association during the COVID-19 pandemic.

3. Methodology

3.1. Sample Selection

All firms listed between 2015 and 2021 on GCC (Saudi Arabia, United Arab Emirates, Bahrain, Qatar, Oman, and Kuwait) bourses comprised this study’s sample. The whole sample utilized in this study is shown in

Table 1. Study sample.

Sector	Included	Excluded	Final Sample	%
Services	201	3	200	66%
Industry	102	0	100	34%
Total	303	3	300	100%

. Information on CSR expenditure and activities, as well as CS proxy, was acquired via Bloomberg, Refinitiv, and Yahoo. Some information, such as that pertaining to control variables, was also manually collected from the annual reports of the corporations. Furthermore, a few firms were excluded from the sample because they had closed, were suspended from the exchange, or lacked CSR information. The total number of firms included in the current study was 300 and the total number of observations was 2100.

3.2. Measuring CSR and CS

The subsequent sections discuss how the dependent variables are measured utilizing the EM method of CS. In addition, a detailed explanation of the independent variable (i.e., CSR) measurements and control variables is given below. The independent variable, dependent variable, and control variables were measured according to prior studies (McVay 2006; Zalata et al. 2019).

3.2.1. CSR Measurement

This study employed Bloomberg’s environmental, social, and governmental disclosure score as a proxy for the degree of CSR disclosure, in keeping with earlier research (e.g., Giannarakis 2014; Nollet et al. 2016; Ghardallou and Alessa 2022). Hazardous waste, environmental operating rules, total greenhouse gas emissions, renewable energy, emissions, materials, and water pollution are just a few of the elements used in Bloomberg’s environmental statistics. Bloomberg calculates the social disclosure score using employee turnover, workplace accidents, overall fatalities, the employment of women, and community impact policy information. Governance particulars consist of executive compensation, board structure and function, business participation in politics, and board committees. Bloomberg provides a comprehensive CSR proxy that is rated from 0 to 100. Firms that disclose the minimum requirement will score 0.1, while firms will score 100 if they disclose all the required data. The score is also modified for different sectors. According to Giannarakis et al. (2011) and Giannarakis (2014), every industry sector has different CSR issues and features; hence, each firm is only assessed using the statistics that pertain to their sector.

3.2.2. CS Measurement

This study focuses on the misclassification of recurrent costs as proposed by McVay (2006). It examines the link between unanticipated core profits and non-recurring expenses to investigate the relationship between CSR and CS factors. Fan et al. (2010) suggested altering McVay’s (2006) model by removing the accruals variables for the current year. Thus, in accordance with other research, a modified form of McVay’s (2006) model (e.g., Zalata et al. 2019) was employed in this investigation.

The first stage in assessing the CS technique is to cross-sectionally estimate a proxy for the typical/expected core earnings using the following model:

Equation (1):

$${CE}_{i,t}={\beta }_0+{\beta }_1{CE}_{i,t-1}+{\beta }_2{ATO}_{i,t}+{\beta }_3{ACCRUALS}_{i,t-1}+{\beta }_4\mathsf{\Delta }{SALES}_{i,t}+{\beta }_5{NEG\_\mathsf{\Delta }SALES}_{i,t}+{\epsilon }_{i,t}$$

(1)

where ${CE}_{i,t}$ represents core profits prior to depletion and non-recurring items. $CEi,t-1$ represents the core earnings from the prior year. $ATO(i,t)$: the sales divided by the average net operating assets is the asset turnover ratio. $ACCRUALS(i,t-1)$: operating accruals are assessed as net income before exceptional items, such as operating cash flow/sales; operating accruals are computed as prior-year operating accruals. $ACCRUALS(i,t-1)$ represents the accruals from prior periods. The percentage change in sales is represented by ΔSALES(i,t) = $Sales\left(t\right)-Salest\_(t-1))/Sales(t-1)$. $NEG\backslash SALES(i,t)$: the percentage change in sales ΔSales(i,t) is equal to 0 in the case of a positive change in sales, and it takes a different value in the case of a negative change in sales.

Unexpected or uncommon core earnings (UCEs), which are the disparities between reported core earnings and expected core earnings, were then determined after estimating the ordinary/expected core earnings using the coefficients from Equation (1). If the unexpected core earnings were positive, the company probably converted regular expenses to non-recurring expenditures to increase core earnings.

The current study examined whether the sample businesses misclassified recurring expenditures as non-recurring (NREC) in a way similar to Orjinta (2018) and Zalata and Roberts (2016). Research has also confirmed that IFRS non-recurring items provide companies greater leeway to use CS (Anagnostopoulou et al. 2021).

As indicated by Anagnostopoulou et al. (2021), the variance between reported core profits and net income represents non-recurring items. Anagnostopoulou et al. (2021) specified that a company is considered a classification shifter if its unanticipated core profits are positive and its non-recurring expenses decrease its income; if not, it receives a zero rating. As a result, the CS proxy may be zero or positive. Positive scores show the number of recurrent expenses that are incorrectly categorized as non-recurring earnings-decreasing expenditures for CS, which signifies engagement in this strategy (Joo and Chamberlain 2017).

Regression model:

In order to evaluate the hypothesis, the regression model below was constructed with CSR as an independent variable. Additionally, as indicated by previous research (Waddock and Graves 1997), control variables were included in the model, as shown in

Table 2. Study variables description.

Label	Variable Name	Measurement
CS-related variables
UCE	Unexpected core earnings	The difference between reported earnings and expected core earnings using Equation (1) scaled by sales.
NREC	Non-recurring expenses	The difference between reported core earnings and bottom-line earnings scaled by sales (positive differences correspond to income-decreasing items, whereas negative differences correspond to income-increasing items and are set to zero).
Independent variable
CSR	Corporate social responsibility	Environment, social, and governance performance scores collected from the Bloomberg database. The score ranges from 0 to 100.
Control variables
ROA	Return on assets %	Net income/average total assets
LFSZ	Firm size	Natural logarithm of total assets
LVG	Leverage%	Total liabilities/total assets
AGE	Firm age	The difference between the establishment date of the firm and the report date.
COVID	COVID-19 period	COVID-19 period indicator variable is set to one if firm-year observation is after 2020 and zero otherwise.

. This study examined the relationship between unexpected core earnings and non-recurring expenditures in order to refute the assertion made by McVay (2006) that recurring expenses are commonly misclassified. In order to examine the relationship between non-recurring costs (NREC) and the CSR variable (NREC × CSR), this study included NREC as an interaction variable, in accordance with Orjinta (2018) and Zalata and Roberts (2016). The research indicates that the connection should be substantial. The difference between real core profits and net income scaled by sales was used to calculate the NREC. The following describes the model that was used to examine the relationship between CSR and CS:

Model (2):

$$UCE={\beta }_0+{\beta }_1{NREC}_{it}+{{\beta }_{2}CSR}_{it}+{\beta }_3{NREC}_{it}\times {CSR}_{it}+{\sum }_{k=1}^4{\beta }_i{Control}_{it}+{\mu }_i+{{\eta }_t+\mathrm{ꜫ}}_{it}$$

(2)

where:

• UCE: Unexpected core earnings.
• NREC: Non-recurring expenditure.
• CSR: A proxy measured using Bloomberg environmental, social and governance score. Control: A set of control variables related to firms’ characteristics discussed in Table 2.
• ${\mu }_i$: Firm fixed effects.
• ${\eta }_t$: Year fixed effect.
• ${\mathrm{ꜫ}}_{it}$: The error term.

The researchers incorporated a COVID proxy for the COVID-19 era in Model (2) to investigate whether the COVID-19 pandemic affects the relationship between a company’s CSR and CS. In the event that the firm-year observation was made after 2020, a dummy variable named COVID had a value of one; otherwise, it had a value of zero. The regression model is as follows:

Model (3):

$$\begin{gathered} UCE={\beta }_0+{\beta }_1{NREC}_{it}+{{\beta }_{2}CSR}_{it}+{\beta }_3{COVID}_{it}+{\beta }_4{NREC}_{it}\times {CSR}_{it}+{\beta }_5{NREC}_{it}\times {COVID}_{it} \\ +{\beta }_6{CSR}_{it}\times {COVID}_{it}+{\beta }_7{NREC}_{it}\times {CSR}_{it}\times {COVID}_{it}+{\sum }_{k=1}^4{\beta }_i{Control}_{it}+{\mu }_i+{{\eta }_t+\mathrm{ꜫ}}_{it} \end{gathered}$$

(3)

To investigate the study hypothesis, this study focuses on the coefficients of ${\beta }_1$, ${\beta }_4$, ${\beta }_5$, and ${\beta }_7$.

4. Result and Discussion

4.1. Descriptive Analysis

The descriptive statistics, including the mean, standard deviation, and minimum and maximum values of the variables, are displayed in

Table 3. Descriptive statistics.

Variable	n	Min.	Max	Mean	SD
CSR	2100	2.20	51.822	38.910	4.628
UCE	2100	−1.926	2.344	0.006	3.695
NREC	2100	0	0.754	0.029	0.233
Age	2100	7	61	31.329	9.336
Leverage	2100	0.009	0.872	0.452	0.016
Assets	2100	4.297	838,992	603,925	35,092
ROA	2100	−85.10	66.12	2.1751	15.374

. A comparatively low amount of CSR disclosure is shown by the mean value of the CSR proxy, which is 38.910. The variables are also skewed, according to the findings of the normalcy distribution test. Hence, the natural logarithm is utilized in the regression analysis. The expectation model’s residuals are also reflected by the mean of the unexpected core earnings (UCEs), which is around 0 (0.006). Regarding the control variables, the leverage mean is approximately 452, and the ROA mean is 2.175. The lowest ROA is −85.10, which indicates that some companies included in the study sample face operational or financial challenges. The age distribution of the enterprises is 7–61, with a mean of 31.

Various tests were conducted to determine the multicollinearity, normality, linearity, and autocorrelation of the models and study variables.

Table 4. Pearson’s correlation.

	ROA	Lev.	CSR	Assets	Age	UCE	NREC	COVID
ROA	1
Leverage	−0.044	1
CSR	−0.035	−0.212 **	1
Assets	0.107	0.190	0.356	1
Age	0.047	0.044	−0.067	−0.034	1
UCE	0.145 **	0.158 *	0.363	0.047	0.081	1
NREC	−0.056	0.083	0.028	0.082	0.013	−0.438 **	1
COVID	0.140	0.296	0.021	0.046	0.005	0.0823	0.005	1

** Correlation is significant at the 1% level. * Correlation is significant at the 5% level.

shows the Pearson’s correlation results. Since no coefficient in the current study was bigger than 80, multicollinearity was generally not an issue.

The variance inflation factor (VIF) was also used to test for multicollinearity. The VIF coefficients are shown in

Table 5. Collinearity and normality statistics.

Variables	Tolerance	VIF	Skewness	Kurtosis
CSR	0.325	4.631	15.207	352.621
ROA	0.811	3.624	0.721	−2.201
Assets	0.478	3.415	10.221	281.240
LVG	0.379	2.871	0.283	2.378
AGE	0.755	2.995	0.221	−1.516
COVID	0.821	4.011	0.433	−1.942

, and the scores in the three models did not exceed 10. Therefore, multicollinearity was not a problem.

Table 5 also demonstrates that all study variables, with the exception of CSR and firm size, had normal distributions according to the results of the skewness and kurtosis tests. These two variables were therefore converted to natural logarithms. The autocorrelation test (Durbin–Watson) is presented in

Table 6. The relationship between CSR and CS.

Variables	Model (1)		Model (2)		Model (3)
	Coef.	t-Statistic	Coef.	t-Statistic	Coef.	t-Statistic
NREC	0.082 *	1.971	0.047 *	1.939	0.054 *	1.879
CSR	−0.002	−0.229	−0.006	−0.954	−0.008	−0.412
COVID			0.197 ***	3.521	0.325 ***	3.284
NREC × CSR	−0.290 ***	−5.233	−0.210 ***	−2.915	−0.328 ***	−2.108
NREC × COVID			0.199 ***	3.517	0.221 ***	5.130
CSR × COVID					0.002	0.821
NREC × CSR × COVID					0.112 ***	3.826
ROA	0.004	0.231	0.004	0.228	0.004	0.227
ASSETS	0.057 *	1.970	0.072 *	1.992	0.061 *	1.982
LEVERAGE	0.461 ***	3.256	0.721 ***	2.192	0.290 ***	3.512
AGE	0.023	0.543	0.053	0.812	0.013	0.751
CONSTANT	0.015	1.506	0.032	1.428	0.032	1.521
Firm fixed effect	Yes		Yes		Yes
Year fixed effect	Yes		Yes		Yes
Regression F	22.091 ***		21.191 ***		24.196 ***
$\mathrm{Adj}.{\mathrm{R}}^2$	0.422		0.451		0.437
D-W	2.259		2.121		2.849

***, * represent significance at the 1%,and 10% levels, respectively.

, and, according to the results, none of the three models had an issue with autocorrelation because the (D-W) value was nearly equal to 2.

4.2. Multivariate Analysis

To examine the link between CSR and CS, we focus on the NREC coefficient and the interaction between NREC and CSR. NREC is positively and strongly associated at the 10% level, according to the results for Model (1) in Table 6. This suggests that certain core items may have been mistakenly categorized by firms as non-recurring items in the income statement, inflating core profits. The substantial negative value of 1% for the relevant variable NREC × CSR indicates that enterprises that participate in CSR activities are more likely to eliminate CS as an EM strategy. The findings support the agency hypothesis (Jensen and Meckling 1976; Dechow et al. 1995), which was proposed in earlier research, and they indicate that companies should participate in CSR initiatives to enhance their reputation among stakeholders. However, they may cover their immoral or opportunistic behavior in order to profit privately (Choi et al. 2013; Mohmed et al. 2020). H1 is therefore accepted.

However, because prior research was based on the full sample period, it overlooked the potential that businesses’ opinions regarding EM would alter after the COVID-19 era. To test the second hypothesis, we thus extended the study to take into consideration the impact of the COVID-19 pandemic on the link between CSR and CS. An interaction term that included COVID, NREC, and CSR was developed in order to test this. The results for Models (2) and (3) are listed in Table 6. The results for Model (2) show that, during the COVID-19 period, companies’ preference for the CS approach increased, as NREC is positive and significant at 10%, and NREC × COVID is positively significant at 1%.

Furthermore, the findings for before and after the COVID-19 era are displayed in Model (3) in Table 6. At 10%, NREC remains statistically significant and positive, indicating that businesses used CS. Furthermore, the data show that NREC × COVID is positive and significant at 1%, indicating that CSR-engaged businesses used CS more frequently, especially following the COVID-19 period. Interestingly, the coefficient for NREC × CSR is significantly negative at 1%, suggesting that prior to the COVID-19 era, companies that engaged in CSR activities used CS less frequently.

Model 3 in Table 6 shows that NREC × CSR × COVID is significantly positive at 1%, suggesting that CS gained prominence in companies that participate in CSR activities following the COVID-19 era. This implies that the COVID-19 timeframe influenced the degree of preference that CSR-engaged enterprises have for CS. In other words, following the COVID-19 era, corporations that engaged in CSR efforts were more likely to misclassify the income statement items as an EM strategy. In contrast, they were less likely to do so prior to that time. The results show that corporations resort to another EM approach, such as CS, which has a lower detection cost, when they are under pressure to enhance their financial performance. The findings support the claims made by Ozili (2020) and Carracedo et al. (2021), which state that it is difficult for enterprises to both survive and improve their financial performance during trying times such as pandemics. Because of this, companies are more inclined to use EM in times of crisis or epidemics. H2 is therefore accepted.

Our results confirm the stakeholder theory’s suggestion, which is that companies that performed CSR activities before the pandemic are less likely to engage in the CS technique. This study’s results, however, demonstrate that this relationship became significantly positive after the outbreak of the pandemic. This suggests that companies’ CSR efforts may have been utilized to cover their opportunistic CS actions during that hard time (agency theory). Overall, the results show that the pandemic caused businesses to change their values. For example, even though CS is believed to be less risky than the other EM approaches, it is still unethical to carry out more CSR activities to hide the growing use of CS mechanisms during challenging situations such as a pandemic.

4.3. Robustness Tests

The current study updated the McVay (2006) model initially based on recommendations by Fan et al. (2010). To prevent biased outcomes in the McVay (2006) model, several researchers asserted that total accruals should be substituted for working capital accruals by excluding depreciation expenditures and other non-recurring components (Anagnostopoulou et al. 2021). As a result, in a robustness evaluation, working capital accruals were substituted for total accruals to re-estimate the UCE.

Table 7. Alternative CS measurement.

Variables	Model (1)		Model (2)		Model (3)
	Coef.	t-Statistic	Coef.	t-Statistic	Coef.	t-Statistic
NREC	0.049 *	1.910	0.035 *	1.982	0.067 *	1.981
CSR	−0.005	−0.522	−0.002	−0.600	−0.003	−0.412
COVID			0.132 ***	2.980	0.172 ***	4.220
NREC × CSR	−0.213 ***	−2.510	−0.116 ***	−2.962	−0.192 ***	−3.459
NREC × COVID			0.436 ***	4.520	0.190 ***	3.181
CSR × COVID					0.011	0.906
NREC × CSR × COVID					0.139 **	2.099
ROA	0.011	0.619	0.012	0.623	0.011	0.612
ASSETS	0.028 *	1.998	0.045 *	1.970	0.095 *	1.969
LEVERAGE	0.219 ***	3.756	0.128 **	2.510	0.388 ***	3.933
AGE	0.046	0.595	0.000	0.529	0.002	0.661
CONSTANT	0.052	1.670	0.050	1.492	0.026	1.539
Firm fixed effect	Yes		Yes		Yes
Year fixed effect	Yes		Yes		Yes
Regression F	29.012 ***		26.295 ***		27.231 ***
$\mathrm{Adj}.{\mathrm{R}}^2$	0.423		0.460		0.422
D-W	2.912		2.850		2.622

***, **, * represent significance at the 1%, 5%, and 10% levels, respectively.

provides an example of the findings. The outcomes confirm the conclusions made in the primary analysis and are consistent with the primary CS regression model that was used.

The possibility of reverse causation between CS and CSR variables, often known as the simultaneity issue, is another issue that has to be addressed. Controlling simultaneity, a part of the endogeneity problem, is necessary to avoid bias in the main regression results. As shown in

Table 8. System GMM regression analysis results.

Variables	Model (1)		Model (2)		Model (3)
	Coef.	t-Statistic	Coef.	t-Statistic	Coef.	t-Statistic
LAG_UCE	0.521 ***	4.129	0.309 ***	3.699	0.226 ***	2.985
NREC	0.076 *	−1.989	0.082 *	1.896	0.066 *	1.971
CSR	−0.065	−0.321	−0.006	−0.598	−0.032	−0.425
COVID			0.054 ***	3.622	0.321 ***	3.662
NREC × CSR	−0.262 ***	−3.712	−0.329 ***	−3.210	−0.122 ***	−4.521
NREC × COVID			0.212 ***	2.891	0.262 ***	4.143
CSR × COVID					0.009	0.342
NREC × CSR × COVID					0.124 **	2.172
ROA	0.001	0.221	0.001	0.225	0.008	0.205
ASSETS	0.087 *	1.919	0.152 **	2.102	0.079 *	1.891
LEVERAGE	0.169 **	2.240	0.189 **	2.173	0.193 **	2.287
AGE	0.015	0.402	0.212	0.421	0.021	0.412
INTERCEPT	0.010	1.239	0.021	1.498	0.010	1.492
Firm fixed effect	Included		Included		Included
Year fixed effect	Included		Included		Included
AR (1) (p-value)	0.000		0.001		0.000
AR (2) (p-value)	0.742		0.654		0.523
Sargan test (p-value)	0.000		0.000		0.000
Hansen test (p-value)	0.286		0.269		0.245

***, **, and * represent significance at the 1%, 5%, and 10% levels, respectively.

, the regression was re-estimated using the system GMM regression technique to verify the robustness of the main regression, as the use of this technique is suggested by both Arellano and Bover (1995) and Blundell and Bond (1998) for reducing concerns regarding reverse causality and unobserved variables that may be related to CS and CSR variables (e.g., Nekhili et al. 2020). In accounting research, this technique is frequently employed to address the endogeneity problem (Gull et al. 2018; Zalata et al. 2019).

5. Conclusions

Since businesses are becoming more concerned with their moral and ethical behavior (Waheed and Zhang 2022), the CSR concept has drawn much attention and has become one of the most hotly debated academic issues. According to agency theory, researchers have argued that companies may participate in CSR initiatives to cover up their opportunistic practices, such as EM practices. However, many researchers agree that, in accordance with the stakeholders’ theory, businesses that actively engage in CSR initiatives simply have higher ethical standards, which cause them to view EM practices as unethical and self-serving. The majority of studies conducted thus far have concentrated on analyzing the connection between the AM and a particular type of EM. Nevertheless, no study has found a connection between CSR and CS, a less expensive EM technique. The literature’s assertion that the reason for utilizing CS is comparable to that of other approaches (such as the AM and RM), which is to misinform stakeholders, gives rise to the necessity of testing this link.

Moreover, earlier research has demonstrated that managers have recently been employing this strategy more frequently than the accrual method. The fact that auditors and regulators have limited capabilities to identify its use makes it less risky (Alfonso et al. 2015). Additionally, authorities have recently started to pay greater attention to the reporting of core profits, as this has grown in popularity as a performance statistic in financial markets. Furthermore, even though the CS method is referred to as a “soft EM method” and is less risky, it could have severe repercussions because it could misinform investors about a company’s future performance. Similar to the AM, controlling core profits can also negatively impact future operational outcomes and cash flows. Therefore, in contrast to other investigations, this study offers a more comprehensive understanding of the link between CSR and EM by investigating the CS approach as a less risky EM method.

The COVID-19 pandemic increased the number of internal and external CSR activities, which raises another important point: did businesses use the pandemic to hide their opportunistic practices by increasing their CSR activities? It is crucial to determine how the pandemic impacted this relationship because it is unclear how businesses would act morally in crises such as a pandemic, where EM methods with lower detection costs, such as CS, can be used. This study thus provides an answer to this important empirical research question. The idea that, during times of crisis or a pandemic, firms are more likely to participate in CSR activities in support of their EM practices may be tested in a fresh empirical context with COVID-19.

We use a sample of GCC-listed firms from 2015 to 2021 to examine the relationship between CSR activities and the CS technique. Our results confirm the stakeholder theory’s assertion that firms with higher levels of CSR before the pandemic are more likely to have lower levels of CS practices. However, this study’s results indicate that this association was considerably positive during the pandemic, indicating that corporations’ CSR initiatives could have been covering for their EM practices during that trying time (agency theory). According to the findings, the pandemic has affected businesses in terms of changing their values. For example, while CS is perceived as being less risky than the other EM methods, it is still unethical to engage in more CSR activities in order to hide the growing use of CS activities during a trying time such as a pandemic.

This study’s conclusions are useful to investors, boards of directors, regulators, auditors, and policymakers who are concerned with CSR activities and profit quality. The current study is sound; however, its limits should be made clear. Given that this study only examined a portion of GCC businesses, it is crucial to think about increasing the sample size. Second, this study employed the CS technique proposed by McVay (2006). Future research, however, may make use of alternative models, such as the one suggested by Anagnostopoulou et al. (2021), that are capable of measuring CS practices. Furthermore, although this study focused on firms’ behavior during the pandemic, the results can be generalized, as the pandemic represents a crisis and a tough time for firms. Hence, future studies could use similar scenarios and different crises to check whether firms’ values change during a difficult time. Furthermore, business loss and auditor quality may be included as control variables in models in future research.

Given the laxer regulations pertaining to non-recurring items in the income statement and the limited oversight function of auditors with regard to CS methods, the bulk of the literature has demonstrated that CS aims to misinform stakeholders and may have negative effects (Zalata and Roberts 2016). Despite this, CS may not seem like a serious EM issue. As a result, efforts to eradicate CS behaviors—such as reducing the degree of flexibility in income statement item classification under IFRS—need to be intensified. In order to possibly avoid inaccurate income statement item classification, regulators should also implement new regulations. In addition, since CS is extensively used and has the potential to mislead stakeholders—especially in times of crises such as a pandemic—serious legal action must be taken to eradicate it. Finally, while several tests were conducted to assess the validity of this study’s conclusions, future research could make use of additional econometric techniques or instrumental variables to address problems with endogeneity and heterogeneity in the data and enhance the current analysis.