Factors Affecting CSR Disclosure by Takaful Insurance Companies During the Pandemic Crisis

Abstract

This study explores the key factors driving corporate social responsibility disclosure (CSR_DISC) by Takaful insurance companies (TKIs) in Saudi Arabia during and after the COVID-19 pandemic. We use content analysis and follow an unweighted scoring method to score the CSR_DISC index. Based on a sample of 26 Saudi-listed TKIs, for the period 2020–2024, we employ Poisson panel and negative binomial panel models to examine the interdependent relationships between CSR_DISCs and a set of corporate governance factors. We find that Saudi TKIs increased their CSR_DISCs in their financial reporting during and after the COVID-19 crisis. These findings confirm that board and firm size have a significant and negative effect on corporate CSR_DISC. However, the number of independent board members and female directors positively affect the extent of CSR_DISCs. Finally, the size of the audit committee and the Shariah supervisory board, frequency of board meetings, and profitability do not affect CSR_DISCs.

1. Introduction

Corporate social responsibility (CSR) is related to societal and ethical standards (Mitnick et al., 2023). CSR is a critical success factor for businesses seeking to enhance their reputation, strengthen competitive advantages, increase long-term value, and benefit from tax incentives (Farook et al., 2011).

Because of the growing importance of CSR_DISC in business, numerous studies, drawing on multiple theoretical perspectives, have explored its determinants (Harun et al., 2020; Hussain et al., 2018; Habbash, 2016). Legitimacy theory asserts that organizations consistently strive to operate within the boundaries and norms of their respective societies, aiming to ensure that their actions are regarded as legitimate by external stakeholders (Deegan, 2002). As for this theory, companies reveal their CSR activities to show that operations are legitimate and that they are decent corporate citizens (Cho & Patten, 2007). Stakeholder theory focuses on how an organization interacts with particular stakeholders (Freeman, 1984). Conforming to this theory, companies convey their accountability to stakeholders through CSR reporting and actions. Agency theory is a concept describing the relationship between principals (i.e., shareholders) and agents (i.e., directors), where the principals delegate decision-making authority to the agents, who are responsible for managing the organization on their behalf (M. Jensen & Meckling, 1976). Consistent with this theory, corporations are encouraged to disclose more CSR practices in order to mitigate information asymmetry and, consequently, lower agency costs.

Since the unprecedented COVID-19 outbreak, this self-regulating business model has become more important for building trust and integrity with investors because of the environmental and social burdens, as well as uncertainties, faced by companies. In this context, CSR takes on its full meaning and represents a solution that can mitigate the consequences of catastrophic events (García-Sánchez & García-Sánchez, 2020). Hence, during periods of high risk and uncertainty, CSR_DISC is considered extremely important by risk-averse and ambiguity-seeking investors, who are cautious about their investment decisions. They demand a high level of CSR_DISCs to become confident about investing in a firm (Su et al., 2023). Thus, companies disclosed their CSR practices during the pandemic to build trust with stakeholders and mitigate risks arising from a high level of uncertainty. This argument is sustained by scholars who have shown that companies increase their CSR_DISCs during periods of adversity, for example, the COVID-19 pandemic (Bahadar & Zaman, 2022; Ding et al., 2021; Albitar et al., 2021).

There is no agreement in the literature on the extent of companies’ disclosure of CSR practices during periods of high uncertainty. Certain scientific studies indicate a negative or nonsignificant relationship between CSR_DISCs and high levels of uncertainty. They argue that during periods of uncertainty, companies are faced with financial constraints that affect CSR reporting (Ding et al., 2021). Companies limit the disclosure of such information during a crisis because CSR-related information disclosure requires resources and involves costs.

Managing risk and uncertainty is a key function of insurance businesses and an essential element of risk management and accountability (Hemrit & Belgacem, 2024). Insurance companies are pivotal for combating crises and implementing societal responsibility principles in precarious situations. CSR_DISC by insurance companies was a valuable instrument for managing and gaining the trust of stakeholders during the COVID-19 pandemic (Williams, 2024). Okhrimenko and Manaienko (2019) show that insurers gain competitive and profitability advantages if they are seen to be socially responsible and disclose sufficient information. Poursoleyman et al. (2024) demonstrate that sustainable investments and accompanying reports improve companies’ resilience in the face of environmental and social crises, thereby lowering investment risks. Khalaf and Nobanee (2022) assert that an increase in CSR_DISCs serves as a sign of progress and long-term success within the insurance sector.

This research enhances the existing literature on CSR_DISC in several ways. First, this study is, to our knowledge, the first empirical investigation of CSR_DISCs by TKIs in Saudi Arabia during and subsequent to the COVID-19 outbreak. The main concentration on Saudi Arabia’s insurance sector is because the KSA has the most extensive insurance market among the Gulf Cooperation Council (GCC) nations. In addition, CSR is an essential component of Saudi Arabia’s Vision 2030, which states that Takaful insurance companies (TKIs) should benefit both participants (i.e., the insured) and society. Furthermore, regulations for corporate disclosure in Saudi Arabia have evolved considerably in recent years. In 2017, the Saudi government implemented reforms targeting corporate governance regulations, which mandated that the mentioned corporations must provide non-financial data related to their CSR operations. In 2023, Saudi Arabia implemented international Financial Reporting Standards (IFRS 17) to enhance the quality of financial reporting, necessitating significant additional disclosures (SAMA, 2023). Second, the literature (Kraus, 2024; Ridwan & Mayapada, 2022; Habbash, 2016) focuses on the determinants of CSR_DISC in an economically stable framework characterized by certainty. This research concentrates on the factors contributing to CSR_DISC during and after the COVID-19 pandemic within an uncertainty framework. Third, only a small number of research studies examine the connection between Islamic business governance (ICG) factors and CSR_DISC based on shariah principles, but they are limited to the banking industry (Jan et al., 2021; Amran et al., 2017). Investigating this relationship in the insurance sector will enrich the literature. Fourth, most studies focus exclusively on panel modeling with generalization, which undermines the very idea of sustainability. Considering the specific characteristics of the disclosure index, this study is unique in examining the factors contributing to CSR_DISC using count models, including Poisson and negative binomial panel models. Drawing on a sample of 26 Saudi-listed TKIs, for the period 2020–2024, our findings reveal that Saudi TKIs increased their CSR_DISCs in financial reporting during and after the COVID-19 crisis. The results indicate that both board size and firm size have a significant negative impact on CSR_DISCs. In contrast, a higher number of independent board members and the presence of female directors are positively associated with the extent of CSR_DISCs. Lastly, variables such as audit committee size, Shariah Supervisory Board size, board meeting frequency, and profitability show no significant effect on CSR_DISCs. The organization of the paper is as follows: Section 2 provides an overview of the existing literature on the determinants of CSR_DISCs in an uncertainty framework, as well as the hypothesis statements. Section 3 presents the methodology and data, Section 4 discusses the results, and Section 5 presents the conclusion.

2. Existing Literature and Hypothesis

2.1. CSR_DISC and Uncertainty Caused by the Pandemic

Few theories explain CSR_DISC during a crisis. Legitimacy, stakeholder, and agency theories suggest that CSR protects firm value during a crisis. In line with legitimacy theory, during times of crisis, CSR_DISC can play a pivotal role in safeguarding corporate legitimacy by demonstrating the organization’s commitment to ethical behavior, social responsibility, and transparency (Dowling & Pfeffer, 1975). In situations where a crisis arises, organizations must work to reaffirm their social legitimacy by aligning their behavior with widely accepted community standards. Engaging in CSR initiatives is often viewed as an effective means of achieving this alignment. Empirical studies have supported this view. For instance, Deegan (2002) highlights that corporate social disclosure is largely voluntary and suggests that managers frequently use it as a strategic tool to address perceived challenges to the organization’s legitimacy, especially in times of heightened stakeholder scrutiny and concern. Aguilera et al. (2007), drawing on legitimacy theory, suggest that CSR initiatives allow firms to align with societal values and expectations and that sustaining these efforts during periods of crisis is essential for retaining stakeholder trust and support. Summerhays and De Villiers (2012) apply the legitimacy theory to the analysis of corporate reports, arguing that in the presence of uncertain events, the company tends to increase its CSR_DISC in annual reports. Likewise, stakeholder theory (Freeman, 1984) emphasizes that organizations must consider the interests and expectations of all stakeholders to maintain long-term success and legitimacy. Buallay et al. (2020) argue that in order to meet the expectations of diverse social groups, businesses must engage with stakeholders and understand their needs; maintaining strong relationships with key stakeholders is likely to enhance overall performance. Noviarty and Edryani (2021) emphasize that during a pandemic, companies are expected to be more responsive to societal conditions, and increasing corporate social responsibility (CSR) initiatives is one effective way to fulfill this expectation. Thus, CSR_DISC is a way of communicating with stakeholders during a crisis to reassure them and create a positive impression of a firm, which prevents or decreases the negative consequences of a crisis (Bahadar & Zaman, 2022). In the view of Agency theory (M. Jensen & Meckling, 1976), agents provide corporate performance reports to principals to reduce information asymmetry. Therefore, managers must participate in CSR_DISC to bridge the information gap between stakeholders and organizational leaders, particularly during periods of uncertainty (Wild & Wild, 2023). Elzahar and Hussainey (2012) argue that, in line with agency theory, larger firms are expected to disclose more information to mitigate higher levels of information asymmetry and agency costs. Waris et al. (2017) argue that by providing more transparent CSR information during COVID-19, firms aim to mitigate agency problems, reassure stakeholders, and protect firm value. Al-Hadi et al. (2022), drawing on agency theory, suggest that managers have an incentive to voluntarily disclose CSR activities to reduce agency costs. Enhanced CSR _DISCs during crises like COVID-19 can, thus, be seen as a mechanism that aligns managers’ interests with those of shareholders and broader stakeholders.

COVID-19 has underscored the necessity for forward-looking knowledge regarding its impacts on organizations. The pandemic also forced corporations to demonstrate improved societal and ethical behavior through CSR_DISC (Albitar et al., 2021). It forced investors and portfolio managers to look for ways to manage risk during uncertain times more effectively. Companies with a high level of CSR involvement inspire more trust in investors. Therefore, CSR can be used strategically to enhance the clarity of investment data to decrease volatility, such as a stock price crash (Karagiannopoulou et al., 2023). Moreover, the ambiguity over the COVID-19 outbreak affected investor state and incited panic and anxiety between traders. Adverse sentiment adversely impacts asset valuation and the efficiency of the stock market. Consequently, one way in which companies can protect themselves against deteriorating investor sentiment is by investing in CSR activities and revealing these actions in their reports to provide an optimistic image to stakeholders (Ding et al., 2021).

2.2. CSR_DISC and Takaful Insurance

Takaful is a system that integrates the principles of mutual assistance (ta’awun) and voluntary contribution (tabarru), where participants collectively and voluntarily share risk. It also incorporates elements of equity-based financing and profit-sharing, functioning similarly to a conventional stock company in terms of capital growth and returns (Hemrit, 2020).

By adhering to Islamic religious principles, TKIs are presumed to participate in CSR activities that enhance both social welfare and economic progress (Brogi et al., 2022). With TKIs, fellow Muslims who suffer a misfortune are helped through voluntary contributions to a shared fund, or tabarru’, which is used to withstand the monetary dangers of every insurer. Given that TKIs focus on risk sharing rather than risk transfer, the idea of common prosperity, in which all fellows in the insurance sector share risks, is emphasized. Therefore, TKIs can be expected to be more engaged with society than other companies and disclose more information on their CSR activities. Only a few studies have investigated the CSR_DISCs of TKIs in relation to the COVID-19 pandemic, particularly examining how these companies communicated their social responsibility efforts during the crisis. The literature in this area focuses on Islamic banks, and the few studies that are related to TKIs focus on the regulation of the insurance business regarding CSR practices (Alshammari & Altarturi, 2021; Farook et al., 2011). Another important factor that has been largely neglected by the literature is ICG and its impact on CSR_DISC. ICG is a religious-doctrine-based decision-making process based on the values of responsibility, transparency, independence, justice, discipline, and professionalism; therefore, the existence of an ICG mechanism is anticipated to advantageously influence the level of disclosure (Fadhilah, 2019). Studies on the impact of ICG on CSR_DISC report inconsistent findings. Amran et al. (2017) show that ICG is vital for improved CSR_DISC; it pushes companies to disclose information regarding their contribution to the community, resource use, and protection of the environment. However, Ridwan and Mayapada (2022) show that ICG, exercised through the SSB, has no effect on CSR_DISC. The inconsistent findings about the influence of ICG on CSR_DISC motivates us to review their relationship, especially in TKI.

2.3. Research Hypothesis

2.3.1. Corporate Governance Attributes

▪

Board Size (BSZ)

According to the dependence theory, a large board increases transparency through increased monitoring of the top management, which can affect the level of CSR_DISC. With TKIs, BSZ affects the board’s power to manage and evaluate all transactions conducted by the management to guarantee conformity with Shariah law. Directors should have a strong understanding of Shariah principles, ensuring enhanced openness, ethical conduct, and consideration for stakeholder groups (Alabdullah et al., 2019). Indeed, a substantial board can provide a comprehensive representation of ownership, encouraging directors to disclose information that may be beneficial to stakeholders, including CSR activities. Researchers find that there is an optimistic and significant connection amongst BSZ and the level of CSR_DISC (I. Khan et al., 2019; Alabdullah et al., 2019). Thus, we propose the following hypothesis:

H1. 

BSZ is positively associated with CSR_DISCs.

▪

Non-executive independent members (NEIMS)

According to Agency theory, NEIMS restrict the opportunistic behavior of executives by controlling and monitoring their actions. Simultaneously, independent managers contribute to improving corporate social activities (Haniffa & Cooke, 2002). This outcome aligns with legitimacy theory, which asserts that an increase in the number of NEIMS enhances a company’s reputation and financial viability. Independent directors are inclined to encourage voluntary information reports to stakeholders, as they are not actively engaged in corporate management and their compensation is not tied to the company’s financial success (Katmon et al., 2019). Researchers show that NEIMS positively and significantly affect CSR_DISC (Chau & Gray, 2010; Rouf & Hossan, 2021). Hence, we put forward the following hypothesis:

H2. 

A positive correlation exists between NEIMS and CSR_DISC.

▪

Shariah supervisory board size (SSBS)

One important feature that distinguishes TKI from conventional insurance firms is the existence of SSB within their structure, which provides internal governance to ensure compliance with Islamic law, leading to greater CSR_DISC and information transparency. The functioning of the SSB goes beyond ensuring shariah compliance and encompasses environmental, social, and cultural issues. It can be expected that the SSB influences the management to increase their reporting because of their duty and obligation to Islamic law. Hence, it is assumed that getting necessary and sufficient support from the SSB will stimulate CSR adoption by TKI and affect their disclosure level (Harun et al., 2020). SSB has the power and superiority to guarantee that managers behave in a way that enhances the efficacy of their disclosures (Hemrit & Belgacem, 2024). Thus, the following hypothesis is recommended:

H3. 

SSBS has a positive impact on CSR_DISC.

▪

Female Gender Presence (FGP)

As for stakeholder theory, the presence of female board members can improve a company’s relationships with main stakeholders, emphasizing the protection of their interests, and potentially enhancing CSR_DISC (Issa et al., 2021). Females are more socially oriented than men (I. Khan et al., 2019). Consequently, gender board diversity can increase the autonomy and efficiency of decision making and promote CSR performance. Jia and Zhang (2013) demonstrate that female directors enhance voluntary disclosure and the openness of environmental information reporting. Sial et al. (2018) assert that companies with female board participants are more inclined to take part in charity endeavors, thus exhibiting elevated levels of CSR activities. Qiu et al. (2022) illustrate that female directors contribute to greater confidence in a firm through their improvement in CSR_DISC. Consequently, we propose the following hypothesis:

H4. 

FGP has a positive impact on CSR_DISC.

▪

Audit committee size (ACS)

A strong corporate governance structure heavily relies on the audit team. It safeguards stakeholders’ concerns by guaranteeing reliable financial disclosure, such as robust internal controls supported by vigilant oversight (Fama & Jensen, 1983). The auditing committee can enhance auditor efficacy and diminish information disparities among various stakeholders and management, hence elevating the quality of information in financial statements, encompassing CSR_DISC (Dwekat et al., 2022). In this regard, the audit committee is the main element in the promotion of sustainable standards and transparency among corporations. A large audit committee has a greater ability to solve problems in financial reporting and encourage non-financial information disclosure, including CSR_DISC. There is minimal experimental evidence on the association between ACSZ and CSR_DISC. Several studies, drawing on the agency and resource dependency paradigm, show that a large audit committee brings in diverse skills, experiences, and knowledge, resulting in improved control of CSR _DISC (Al-Shaer & Zaman, 2018; Dwekat et al., 2022). Accordingly, the following hypothesis is proposed:

H5. 

ACS positively affects CSR_DISC.

▪

Board meeting frequency (BMFR)

Board meetings are an important avenue for information procurement and exchange between executive (i.e., internal) and independent (i.e., external) directors (Vafeas, 2003). Frequent board meetings benefit stakeholders by delivering greater information transparency. Consequently, regular board meetings guarantee a company’s implementation of sustainable practices (Kılıç et al., 2020). Therefore, frequent board meetings can be expected to better resolve a company’s social and environmental problems and improve the quality of CSR_DISC. Limited research investigates the effect of BMFR on CSR_DISC. Hu and Loh (2018) and Hussain et al. (2018) find that companies that hold regular board meetings generally disclose more regarding their CSR activities. Nour et al. (2020) affirm that BMFR has a favorable impact on CSR_DISC. Consequently, we proposed the following hypothesis:

H6. 

BMFR has a positive effect on CSR_DISC.

2.3.2. Other Control Variables

▪

Size (SZ)

Firm size influences insurance companies’ capabilities and incentives to manage CSR_DISCs. Generally, large insurance companies disclose more information than small companies (Stawati, 2020). According to legitimacy theory, a large company has greater social and environmental impacts than a small company. Because of the greater scale of activities, stakeholders of large insurance companies will pay more attention to their social programs (Williams, 2024). Moreover, major insurance corporations encounter greater societal pressure than smaller ones. Because of the abundance and intricacy of hazards, major insurance firms exhibit heightened awareness of sustainability issues, perhaps enhancing their drive to refine their corporate social responsibility (CSR) management (Gatzert & Reichel, 2022). In contrast, small companies find it difficult to access resources and, hence, tend to use their limited resources to improve financial performance. Therefore, they are less likely to voluntarily become involved in sustainability practices. Brogi et al. (2022) demonstrate a substantial correlation between business size and sustainability performance. Consequently, we propose the following hypothesis:

H7. 

SZ positively affects CSR_DISC.

▪

Profitability (ROA)

Profitability is characterized as a company’s capacity to generate profits. Highly successful insurance firms may be more adept at executing superior CSR management and achieving elevated levels of CSR disclosure (CSR_DISC). This is associated with the legitimacy perspective, whereby highly profitable companies disclose more social responsibility in their annual reports, which signals their competitiveness and efficient performance. Studies show that there is a great connection between firm profitability and CSR_DISC. Latifah et al. (2019) illustrate that the profitability proportion has a constructive effect on the issuance of sustainability reports. Similarly, Brogi et al. (2022) found that there is a good connection in the insurance industry. Therefore, we proposed the following hypothesis:

H8. 

Profitability positively affects CSR_DISC.

3. Study Framework

3.1. Data and Sample Choice

Data were extracted from all 26 TKI firms based in Saudi Arabia that are registered on the Saudi Tadawul Stock Exchange (www.tadawul.com). A total of 130 firm-year reports were collected from financial records and yearly reports spanning 2020 to 2024. The study period covers the COVID-19 pandemic, crash in oil prices, and the adoption of IFRS 17 by Saudi Arabia in 2023. We used annual reports to investigate the CSR_DISC level for two reasons. First, annual reports are the most commonly used document to analyze corporate social activities in the literature (Lanis & Waller, 2009). Second, using annual reports is justified because of their regularity, credibility, accessibility, and the information they provide.

3.2. Variable Measurement

One dependent variable, six independent variables, and two control variables were considered in this study.

Table 1. Definition and measurements of the variables.

Variable	Definition	Measurement	Sources	Expected Sign
Dependent variable
CSR_DISC	Index of CSR disclosure	The number of CSR-disclosed items in the prepared checklist (Appendix A)	Habbash (2016) Alotaibi (2020) Mousa et al. (2018)
Independent variables
BSZ	Board size	Natural logarithm of the total number of directors on the board	Rao et al. (2012)	+
NEIMS	Non-executive independent members	Ratio of the number of non-executive directors to the total number of directors	Rouf and Hossan (2021)	+
SSBS	Shariaa supervisory board size	The number of members on the Shariaa Board	Ridwan and Mayapada (2022)	+
FGP	Female gender presence	The number of female members on the board committee	Qiu et al. (2022)	+
ACS	Audit committee size	Natural logarithm of the total number of members on the audit committee	Al-Shaer and Zaman (2018)	+
BMFR	Board meeting frequency	Natural logarithm of the frequency of board meetings	Hemrit and Belgacem (2024)	+
Control variables
SZ	Size	Natural logarithm of total assets	Al-Shaer and Zaman (2018)	+
ROA	Profitability	Net profit/total of assets	Brogi et al. (2022)	+

Source: authors’ own creation.

presents the variable definitions and measures.

3.2.1. Dependent Variable

CSR_DISC is taken as the dependent variable, which is developed using manual content analysis by organizing data obtained from the yearly reports of all TKI companies from 2020 to 2024. Content analysis is widely used in CSR_DISC studies (Habbash, 2016; Haniffa & Cooke, 2002). Following Alotaibi (2020), Habbash (2016), Muttakin et al. (2015), and Mousa et al. (2018), we constructed a list of 41 items classified into five subcategories (environmental and related information, employee welfare and related information, community involvement and social information, product quality and safety information, and value-added information) to evaluate the level of CSR_DISC in annual reports. For scoring CSR_DISCs, we used an unweighted scoring method, with each item across the five CSR_DISC categories assigned equal significance (Cooke, 1989). A value of 1 is assigned to an item in the CSR categories if it is disclosed; otherwise, the value is 0. Next, we add the scores of each item for all TKIs in the sample and then added the scores of all of the different categories to obtain the CSR_DISC index.

We tested the reliability of the coded data using Cronbach’s alpha (Cronbach, 1951). The reliability criterion for the research instrument is if the reliability coefficient (α) > 0.6. The findings suggest that with Cronbach’s alpha surpassing the 0.8 threshold, the reliability of the coded data series is ensured. In addition, the validity test yields a positive Pearson correlation coefficient ($\mathrm{r}=0.79$), which indicates that the proposed measurement model is acceptable.

3.2.2. Empirical Models

In our research, the dependent variable is a numerical count that only includes non-negative integers; hence, it is appropriate to use both the Poisson and negative binomial panel models for testing the hypotheses and comparison. The Poisson model is widely used to analyze count data due to its positive skewness and leptokurtic nature. Moreover, as an extension of the Poisson estimation, negative binomial panel models are considered useful for predicting count data. We used the following regression model:

$$\begin{gathered} {\mathrm{CSR\_DISC}}_{\mathrm{it}}=\alpha +{\beta }_1{\mathrm{BSZ}}_{\mathrm{it}}+{\beta }_2{\mathrm{NEIMs}}_{\mathrm{it}}+{\beta }_3{\mathrm{SSBS}}_{\mathrm{it}}+{\beta }_4{\mathrm{FGP}}_{\mathrm{it}}+{\beta }_5{\mathrm{ACS}}_{\mathrm{it}}+ \\ {\beta }_6{\mathrm{BMFR}}_{\mathrm{it}}+ {\beta }_7{\mathrm{SZ}}_{\mathrm{it}}+{\beta }_8{\mathrm{ROA}}_{\mathrm{it}}+{\mu }_{it} \end{gathered}$$

(1)

where i denotes the CSR_DISC index, t denotes the year, ${\beta }_0$ represents a constant, and β_i, i =$\left\{1, 2, \dots 9\right\}$ illustrates the predicted coefficients of the explanatory variables. The independent variables are (BSZ), (NEIMS), (SSBS), (FGP), (ACS), (BMFR), (SZ), and (ROA), while ${\mu }_{it}$ is the standard error.

Numbered lists can be added as follows:

• Poisson panel model

Our count variable, CSR_DISC, produces non-negative integer values. Therefore, the Poisson panel model is suitable in this case. Let CSR_DISC_it = y_it. The expected value of y_it, conditioned on X_it, may be written as follows:

$$E\left(y_{it}|X_{it}\right)=Ƶ(X_{it}, \beta )$$

(2)

where $X_{it}$ is a vector of the explanatory variables; β is a vector of the coefficients associated with X_it, and Ƶ is the Poisson distribution that characterizes the model. The joint probability distribution of the i-th observation (unit) is given as follows:

$$E\left(y_{it}|X_{it}\right)={{\displaystyle \prod }}_{i=1}^T{\displaystyle \frac{\exp \left(-{\lambda }_{it}\right){\left({\lambda }_{it}\right)}^{y_{it}}}{y_{it}!}}$$

(3)

where ${\lambda }_{it}$ = ${\mu }_{i}exp\left(x_{it}^{\prime }\beta \right)$ is the Poisson arrival rate, which is determined by $X_{it}$. λ must be a positive real number, that is λ > 0. This means that any parameter (or combinations of parameters) on which we condition our Poisson distribution must be higher than 0. We impose this restriction using an exponential function based on a linear predictor of the explanatory variables. In the fixed-effects Poisson (FEP) model, all time-invariant TKI characteristics are accounted for by the individual heterogeneity term α. We utilized the conditional maximum likelihood (CML), as developed by Hausman et al. (1984), to estimate the model’s variables. The conditional joint density function (CJDF) of the i-th observation is obtained as follows:

$$\mathrm{E}\left({\mathrm{y}}_{\mathrm{i}1,\dots }{\mathrm{y}}_{\mathrm{iT}}|{{\displaystyle \sum }}_{\mathrm{t}=1}^{\mathrm{T}}{\mathrm{y}}_{\mathrm{it}}\right)={\displaystyle \frac{\left({{\displaystyle \sum }}_{\mathrm{t}=1}^{\mathrm{T}}{\mathrm{y}}_{\mathrm{it}}\right)!}{{\left({{\displaystyle \sum }}_{\mathrm{t}=1}^{\mathrm{T}}{\mathsf{\lambda }}_{\mathrm{it}}\right)}^{{{\displaystyle \sum }}_{\mathrm{t}=1}^{\mathrm{T}}{\mathrm{y}}_{\mathrm{it}}}}}{{\displaystyle \prod }}_{\mathrm{t}=1}^{\mathrm{T}}{\displaystyle \frac{{{\mathsf{\lambda }}_{\mathrm{it}}}^{{\mathrm{y}}_{\mathrm{it}}}}{{\mathrm{y}}_{\mathrm{it}}!}}$$

(4)

where $y_{it}$ and ${{\displaystyle \sum }}_{t=1}^T{y}_{it}$ follow a Poisson distribution. If we add the logarithm of all individuals CJDF values, we drive the following conditional log likelihood:

$$ln\mathrm{L}={{\displaystyle \sum }}_{i=1}^N\left\{ln\left({{\displaystyle \sum }}_{t=1}^T{y}_{it}\right)!-{{\displaystyle \sum }}_{t=1}^T\ln y_{it}!+{{\displaystyle \sum }}_{t=1}^T\left[y_{it}{x}_{it}^{\prime }\beta -y_{it}ln{{\displaystyle \sum }}_{t=1}^{T}exp\left(x_{it}^{\prime }\beta \right)\right]\right\}$$

(5)

Thus, we obtain the estimated parameters, ${\beta }_i$, for the FEP model by solving Equation (6), as follows:

$${{\displaystyle \sum }}_{\mathrm{i}=1}^{\mathrm{N}}{{\displaystyle \sum }}_{\mathrm{t}=1}^{\mathrm{T}}{x}_{it}^{\prime }(y_{it}-{\displaystyle \frac{\left({{\displaystyle \sum }}_{\mathrm{t}=1}^{\mathrm{T}}{\mathrm{y}}_{\mathrm{it}}\right)}{\left({{\displaystyle \sum }}_{\mathrm{t}=1}^{\mathrm{T}}{\mathsf{\lambda }}_{\mathrm{it}}\right)}}{\mathsf{\lambda }}_{\mathrm{it}})$$

(6)

• Negative binomial panel model

Poisson regression has one major limitation, as follows: the variance in the count variable cannot be greater than its mean. A common feature of count data is that the variance exceeds the mean. A negative binomial panel model is a more general model for count data, and it is not constrained by the Poisson assumption regarding the variance of the response variable. Therefore, the fixed-effect negative binomial model (FENB) model is more suitable for improving efficiency whenever overdispersion ($var \left(y_{it}\right)$ > $E\left(y_{it}\right)$) is detected (Cameron & Trivedi, 1990). Considering the FENB model, the CJDF for the i-th observation (Hausman et al., 1984) takes the following form (Hausman et al., 1984):

$$\mathrm{E}\left({\mathrm{y}}_{\mathrm{i}1,\dots }{\mathrm{y}}_{\mathrm{iT}}|{{\displaystyle \sum }}_{\mathrm{t}=1}^{\mathrm{T}}{\mathrm{y}}_{\mathrm{it}}\right)={\displaystyle \frac{\Gamma \left({{\displaystyle \sum }}_{t=1}^T{\lambda }_{it}\right)\Gamma \left({{\displaystyle \sum }}_{t=1}^T{y}_{it}+1\right)}{\Gamma \left({{\displaystyle \sum }}_{t=1}^T{\lambda }_{it}+{{\displaystyle \sum }}_{t=1}^T{y}_{it}\right)}}\times \left[{\displaystyle \prod }_{t=1}^T{\displaystyle \frac{\Gamma \left({\lambda }_{it}+y_{it}\right)}{\Gamma \left({\lambda }_{it}\right)\Gamma \left(y_{it}+1\right)}}\right]$$

(7)

where ${{\displaystyle \sum }}_{t=1}^T{y}_{it}$ $~$ NB $\left({\theta }_i{{\displaystyle \sum }}_{t=1}^T{\lambda }_{it}, \left({\theta }_i{{\displaystyle \sum }}_{t=1}^T{\lambda }_{it}\right)\left(1+{\theta }_i\right)\right), {\theta }_i={\displaystyle \raisebox{1ex}{${\alpha }_i$}\!\left/ \!\raisebox{-1ex}{${\mathsf{{\rm Y}}}_i$}\right.}$, and ${\mathsf{\Gamma }}_i$ is the gamma function. By maximizing the following log-CML function, we obtain the CML estimator of the (FENB) model.

$$\begin{gathered} Ln\mathrm{L}={{\displaystyle \sum }}_{i=1 }^N\left\{ln\mathsf{\Gamma }\left({{\displaystyle \sum }}_{t=1}^T{\lambda }_{it}\right)+ln\mathsf{\Gamma }\left({{\displaystyle \sum }}_{t=1}^T{y}_{it}+1\right)-ln\mathsf{\Gamma } \left({{\displaystyle \sum }}_{t=1}^T{\lambda }_{it}+{{\displaystyle \sum }}_{t=1}^T{y}_{it}\right)+ \\ {{\displaystyle \sum }}_{i=1}^T\left[ln\mathsf{\Gamma }\left({\lambda }_{it}+y_{it}\right)-ln\mathsf{\Gamma } \left({\lambda }_{it}\right)-ln\mathsf{\Gamma }\left(y_{it}+1\right)\right]\right\} \end{gathered}$$

(8)

4. Analysis of the Results

4.1. Data Summary

Table 2. Descriptive statistics.

	Mean	SD	Min	Max
CSR_DISC	18.7538	7.5893	7	38
BSZ	2.0889	0.1946	1.6094	2.3978
NEIMS	1.2502	0.2784	0.6931	1.9459
FGP	0.1769	0.4735	0	2
SSBS	0.2612	0.6668	0	3
ACS	1.2289	0.1889	1.0986	1.9459
BMFR	1.6787	0.3341	0.6931	2.3025
SZ	14.6867	1.8562	12	21
ROA	−0.0185	0.1018	−0.5745	0.0786

Source: authors’ own creation.

provides a summary of the descriptive analysis of our model’s variables. The mean of the CSR_DISC during the period under study is 18.75 and ranges from 7 to 38. This score is higher than the scores obtained by Alotaibi (2020) and Habbash (2016) in Saudi Arabia. This outcome aligns with expectations, considering the substantial improvements in disclosure in Saudi Arabia in recent years, driven by the application of the new mandatory corporate governance regulations in 2017 (CMA, 2020) and the adoption of IFRS 17, in 2023, which requires substantial additional disclosures in financial reports (SAMA, 2023). Since all values are below the threshold of 10, this suggests that multicollinearity does not affect the analysis.

4.2. Goodness-of-Fit Tests

We used the full sample to examine the robustness of the results using both fixed- and random-effects Poisson panel models and negative binomial models. We used the Hausman test to ascertain the optimal model choice. The results, as presented in

Table 3. Estimation results.

	Poisson Panel Model (Fixed Effects)		Poisson Panel Model (Random Effects)		Negative Binomial Panel Model (Random Effects)
Variable	Coeff	SD	Coeff	SD	Coeff	SD
Constant	3.0312 ***	0.6795	3.1112 ***	0.7073	604.3761 **	308.5524
BSZ	−0.1619 ***	0.0655	−0.2254 ***	0.1042	−0.4418 **	0.2238
NEIMS	0.0514	0.1403	0.0739	0.1299	0.1915 *	0.1079
FGP	0.3362 ***	0.1085	0.1807 *	0.0941	0.3014 ***	0.0926
SSBS	0.0734	0.0532	−0.0098	0.1078	−0.0098	0.1078
ACS	0.2381 *	0.1435	0.30079 ***	0.1050	0.1407	0.1370
BMFR	0.0069	0.1144	0.0207	0.1055	0.1277	0.1055
SZ	0.0962 ***	0.0435	−0.1018 *	0.0497	−0.3185 ***	0.1440
ROA	−0.0434	0.3158	0.0897	0.3066	0.0836	0.3066
Hausman test (p-value)			34.62 (0.0019)		19.17 (0.2155)
Wald ${\chi }^2$ (p-value)	24.58 (0.0012)		23.32 (0.0092)		19.22 (0.1086)
Overdispersion			53.46 ***

Source: authors’ own creation. Coeff is the estimated value of the coefficient; SD is the standard deviation. *, **, and *** Indicate significance at 10%, 5%, and 1% levels, respectively.

, are in favor of the fixed-effects Poisson model because it is statistically significant. For the negative binomial model, the random-effects model is more appropriate in this study because the obtained ${\chi }^2$ probability is greater than the 5% threshold. Furthermore, the p-value of the Wald test is less than 0.05, which indicates that the two models are statistically significant.

Cameron and Trivedi (1990) emphasize the importance of overdispersion in the analysis of the Poisson panel model fit. The results show that the likelihood ratio test of the overdispersion parameter is 53.46, which confirms the use of the negative binomial model. We discard the null hypothesis, which considers a dispersion parameter of zero. Hence, we conclude that the random-effects negative binomial model is the best fit for the count regression.

4.3. Discussion

Regarding the corporate governance variables, a negative coefficient for BSZ is observed, and it is statistically significant at the 1% level, implying that a large BSZ discourages TKI from disclosing more CSR information. Hence, Hypothesis 1 is not supported. This result corroborates the findings of (M. C. Jensen, 1993), who shows that a large BSZ is susceptible to communication and coordination problems mainly because of the differences in opinions, which makes agreement difficult to achieve. As a result, the board of a TKI becomes ineffective, and CSR practices and their disclosure are ignored. This outcome aligns with that of (I. Khan et al., 2019), who showed that too large a board makes the monitoring process ineffective. An alternative explanation is related to the study period, which is characterized by the detrimental effect of the COVID-19 pandemic on TKIs. Across this period, the board of directors considered that the market was more interested in firm financial performance and not in CSR information.

Although BSZ is found to affect CSR_DISC negatively, NEIMS does not have a significant effect; hence, H2 is also rejected. This may be because during the COVID-19 pandemic, independent board members were more focused on profit and avoiding risky actions that would increase the cost of CSR. Thus, we find that independent directors do not inherently improve board effectiveness, which aligns with the result of Youssef et al. (2019), but this is in contrast with the findings of Qasem et al. (2024) and Habbash (2016).

Moreover, FGP is positive and significant at the $1\%$ level, which confirms H3. This result indicates female directors have a positive part in CSR_DISC. This finding is surprising given that there few women on the boards of Saudi TKI. This result supports the assertion that even if the involvement of female on the board is low, they provide a significant function in enhancing the level of CSR_DISC, because female have been shown to be more cautious in decision making and risk averse. They are more interested in social and sustainability initiatives and more sensitive and ethical than boards composed exclusively of men. This result supports legitimacy and stakeholder theories, according to which female board members send a positive message to stakeholders in terms of meeting society’s expectations and can improve companies’ ability to address multiple stakeholders’ needs through CSR_DISC. Furthermore, the presence of female on the board can support TKIs in handling exogenous shocks, such as the COVID-19 pandemic (Jia & Zhang, 2013). Our findings align with previous studies, which similarly report that diversity positively influences CSR disclosure, whereas larger board sizes are generally associated with a negative impact (e.g., Githaiga & Kosgei, 2023; Le et al., 2023; Di Guida et al., 2022, among others). These results are, in fact, complementary, suggesting that the quality of board composition may be more crucial in shaping CSR strategies than simply the number of board members.

We discovered that there is a negligible and nonsignificant correlation between SSBS and CSR_DISC; thus, H3 is not supported, which is contrary to our expectations. This result can be attributed to the fact that the SSBs of TKIs focus on their duties and responsibilities in the area of operations, such as the approval of new insurance products, ensuring that insurance contracts comply with shariah principles, monitoring of distribution of surpluses, and review of financial statements. Therefore, CSR_DISC in yearly reports is outside of the focus of the SSB of a TKI. As a result, SSBs do not pay more attention to CSR_DISC. Second, most TKIs in Saudi Arabia are cooperative, and having an SSB is optional. It is not mandatory to establish such a board in insurance companies. Thus, the number of TKIs that have an SSB is negligible. In this regard, an SSB has no effect on CSR_DISC. This finding is in line with that of Ridwan and Mayapada (2022), whereas it contradicts the results of Harun et al. (2020), who show that the SSB plays an essential role in CSR_DISC.

The findings indicate that ACS has no substantial effect on CSR_DISC; hence, H5 is rejected. This finding contradicts agency theory, according to which the audit committee is instrumental in monitoring the financial reporting process and ensuring that managers provide transparent disclosure, thereby minimizing information asymmetry and agency problems. The reason for this result may lie in the fact that audit committees are new entities in the Saudi business environment and that non-implementation of the codes does not lead to severe penalties. Additionally, the audit committee’s duties and objectives remain ambiguous. In Saudi Arabia, some companies are unable to identify and outline an audit committee’s functions. This finding supports the conclusions of prior studies (e.g., Habbash, 2016).

Differing from the findings of I. Khan et al. (2019), we find that BMFR and CSR_DISC have a nonsignificant association, which means that the regularity of board meetings has no effect on CSR_DISC. Hence, H6 is rejected. Consequently, the assertion of agency theory concerning the influence of board meeting regularity on CSR_DISC is unsubstantiated. This result is expected given that in the wake of the COVID-19 outbreak, communication and regulatory oversight were weakened because of lockdowns and restrictions on movement, which prevented meetings from being held. Moreover, given the deep economic stagnation caused by the pandemic, many TKIs incurred high losses, and the main focus of the board directors was on improving their poor image. Thus, CSR_DISC was neglected.

Regarding the control variables, we find that SZ is negative and statistically significant at the at 1% level, indicating that large corporations tend to disclose limited CSR practices compared to small businesses. This result is inconsistent with agency theory’s assumption that in large companies, managers mitigate the high information asymmetry and agency costs by providing more disclosure. The influence of COVID-19 on TKIs was remarkable, and even large companies suffered from a decrease in market demand and an increase in market friction, which reduced short-term expenditure, including CSR expenditure. Finally, the results presented in Table 3 indicate that the ROA has no substantial effect on CSR_DISC. This result is expected given the vulnerable state of insurance companies during the pandemic, which was characterized by low profits and high expenses. Hence, CSR activities seem to have been neglected by most TKIs.

4.4. Robustness Test

In order to validate the robustness of the results, we compared the disclosure score to the median. The CSR_DISC score receives a value of 0 for scores below the median, indicating low disclosure, and a value of 1 for scores above the median. Therefore, the dependent variable is binary, assigned values of 1 or 0. In this case, the random-effects logit model as a nonlinear panel model seems to be more appropriate. With a binary outcome, the logistic random-effects model regresses the log odds of the outcome probability on different predictors to estimate the probability of Y = 1, taking random effects into account. The predicted values are the probabilities of CSR_DISC.

Table 4. Logit panel model estimation results.

	Logit Panel Model (Random Effects)		Marginal Effects dy/dx (*)
Variable	Coeff	SD	Coeff	SD
Constant	11.287	12.6529	-	-
BSZ	−12.4082 ***	5.4807	−2.5631 ***	1.2467
NEIMS	0.6238	3.7777	−1.4571	0.9122
FGP	7.4899 ***	4.2507	1.5729 ***	0.4968
SSBS	0.5124	2.0225	−0.3349	0.3167
ACS	9.9363 ***	4.1251	−6.745 ***	1.6904
BMFR	0.0208	2.9452	−0.4417	0.6857
SZ	−1.3195 ***	0.6327	−0.1945 ***	0.0686
ROA	6.5839	16.8398	−5.4628	2.4320
Hausman Test (p-value)	8.22 (0.1728)
Wald ${\chi }^2$ (p-value)	11.77 (0.1616)
Sigma U_rho	87.12 (0.110)

Source: authors’ own creation. Coeff is the estimated value of the coefficient; SD is the standard deviation. *** Indicate significance at the 1% levels, respectively. (*) dy/dx is the discrete change in the dummy variable from 0 to 1.

displays the regression results of the random-effects logit model. The results of our model yield identical conclusions, except for the coefficient of the ACS, which is significant at the 1% level. The results of the Hausman test show that the H = 8.22 and the p-value is 0.1728, which indicate the use of a random-effects logit model. Based on the Hosmer and Lemeshow (2000) test, we find a p-value of 0.1132, indicating no lack of fit.

5. Conclusions

This research investigates the association between ICG factors and CSR_DISC among 26 TKIs in Saudi Arabia during and in the wake of the COVID-19 pandemic. We find that a large BSZ discourages TKIs from disclosing more CSR information in their financial reports. Additionally, we find that the level of CSR_DISCs increases with the number of females in top management but decreases with firm size. Moreover, we find the sizes of the shariah board and the audit board, regularity of board meetings, and profitability do not affect CSR_DISC.

5.1. Implications

The results of this research have significant consequences. First, the execution of corporate governance reforms and the application of IFRS 17 in 2017 and 2023, respectively, by Saudi Arabia have enhanced the CSR practices of TKIs and encouraged additional CSR_DISC. Second, the empirical results reveal that female directors have a significant impact on CSR disclosure. Therefore, we advise regulators to implement policies that promote greater female representation on corporate boards, with the aim of enhancing transparency and sustainability practices. Although FGP plays a more substantial role in CSR_DISC, NEIMS, and ACS, BMFR and SSB do not have a notable impact on CSR_DISC. For instance, regarding the effect of ACS on CSR_DISC, our findings support Institutional theory’s argument that many elements of the corporate governance structure, such as the audit committee, are purely symbolic. Regulators in Saudi Arabia can gain insights from the findings of this study regarding the prominence of the audit committee in the monitoring and supervision of TKI’ financial reporting and disclosure. Additionally, based on the findings of this study, regulators could improve the role of SSBs in TKIs through further regulations that goes beyond Shariah compliance, addressing environmental and social aspects. To enhance the influence of the other corporate governance variables such as BSZ and NEIMS, a closer alignment with CSR goals may improve organizational resilience in time of crises and contribute to building more trustworthy and socially responsible image.

5.2. Limitations

Although this study makes some important contributions, it has the following limitations. First, the CSR_DISC score in this study is exclusively generated from a manual examination of yearly reports, which may yield dissimilar conclusions. A computerized content analysis may enable a study on a larger scale. Second, the sample size is restricted to TKIs in Saudi Arabia, which makes our findings specific to the Saudi context. Future studies should expand the sample size and sectors to validate the results and ensure their generalizability.

Future research could explore several important avenues, including an investigation into the differences in CSR_DISC practices across various financial sectors, such as conventional insurance, banking, and investment companies. This could provide valuable insights into sector-specific strategies and challenges in addressing social responsibility disclosure during crises.