Sustainability assurance quality and board gender diversity: moderating role of assurance level and provider type on the extent of assurance procedures Open Access

Sustainability reports provide stakeholders with information on an organisation’s social, economic and environmental performance (Alkhawaja et al., 2023). The practice developed in response to the complex web of economic, social and environmental challenges facing the world and the acknowledgement that corporate entities have a central role to play in addressing these challenges (Dutta, 2020). The goal is to promote transparency and accountability by providing stakeholders with information on organisation’s sustainability performance and commitment to supporting societal aspirations of sustainable development (Dutta, 2020). While sustainability reporting is now an established global practice ^[1], the credibility of these disclosures continues to raise questions as critics argue that such disclosures are used primarily to facilitate corporate greenwashing (Parguel et al., 2011; Mahoney et al., 2013).

In response, some companies have chosen to secure external/third-party assurance over their disclosures (Farooq and De Villiers, 2020; Free et al., 2024). However, sustainability assurance rates remain low, and have increased slightly in recent years. In 2024, the world’s largest 250 companies had a sustainability assurance rate of only 69%, while N100 companies stood at 54%. This marks a 2% increase for the largest 250 and 5% increase for N100 companies compared to 2017 (KPMG, 2024). However, these trends are likely to change as regulators begin to place greater emphasis on both sustainability reporting, and importantly the assurance of sustainability reports (Farooq et al., 2023). For instance, sustainability reporting is now part of corporate governance codes of many stock exchanges across the globe (KPMG, 2024). Legislators are regulating sustainability reporting which is no longer a voluntary practice in many jurisdictions. For example, the new European Union Corporate Social Reporting Directive, implemented in 2024, will further drive sustainability reporting rates and specifically mandates sustainability assurance for reporters. Further, regulatory agencies are taking punitive actions for incidences of greenwashing indicating their intolerance for greenwashing (Sun, 2023). Sustainability assurance standards are also evolving. For example, in 2024, the International Audit and Assurance Standards Board (IAASB) issued a proposed specialised sustainability assurance standard, International Standards on Sustainability Assurance (ISSA) 5000, to address inadequacies in existing assurance standards, indicating a much more active role for the accounting profession in this field ^[2] with an effective date in 2026.

Sustainability assurance is described as a third-party assurance engagement designed to enhance the credibility of sustainability reports (Farooq and De Villiers, 2019). The aim is to enhance users’ confidence in such disclosures and support efforts towards the attainment of transparency and corporate accountability. Given the importance of sustainability assurance, researchers have sought to examine the quality of sustainability assurance engagements. The concern is that low- or poor-quality sustainability assurance engagements will do little to promote credible reporting, transparency and corporate accountability but rather mislead stakeholders and enhance corporate greenwashing (Farooq et al., 2023).

There remains a dearth of studies examining the factors influencing sustainability assurance quality (Martínez‐Ferrero et al., 2018; Zaman et al., 2021), as well as the impact/consequences of variations in sustainability assurance quality (Farooq and de Villiers, 2019; García‐Sánchez et al., 2019; Khaireddine et al., 2024; Martínez‐Ferrero et al., 2021). There is a need for research focusing on the nature of the assurance work undertaken by assurance practitioners. A performance-based perspective of sustainability assurance quality will provide stakeholders with an understanding of how context specific elements are important in shaping the quality of sustainability assurance based on scope (nature) of assurance procedures employed by practitioners in the assurance process. We thus extend the literature from a performance-based perspective and consider how internal corporate governance attributes, in conjunction with relevant attributes of the assurance process, inform performance-based sustainability assurance quality. Researchers have found that the independence of audit committee members, their industry expertise and meeting attendance rates positively correlate with the quality of sustainability assurance engagements (Zaman et al., 2021) and substantiate the relevance of these factors based on the work undertaken by practitioners during the assurance process. Further, the literature indicates that sustainability assurance quality is higher when the sustainability assurance provider is an accounting firm (Perego, 2009; Martínez‐Ferrero et al., 2018). The type of sustainability assurance provider is an important variable to consider as the market for sustainability assurance remains open to both accounting and non-accounting sustainability assurance providers ^[3]. The type of assurance provider has implications for the work done as the assurance standards, methodologies and overall approach of accounting and non-accounting assurance practitioners can vary considerably.

Our study seeks to examine the association between board gender diversity and sustainability assurance quality. Specifically, we examine the assurance procedures undertaken by practitioners to collect evidence and form an opinion on sustainability reports and highlight the association of these assurance procedures with board gender diversity (BGD). Our use of the number of assurance procedures undertaken as a measure of assurance quality is predicated on the conceptual argument presented by DeAngelo (1981) and Defond and Zhang (2014) regarding the relevance of the auditor’s input and compliance with assurance standards in determining assurance quality. Specifically, the auditor’s input to detecting errors is fundamental to the quality of the assurance process (Rajgopal et al., 2021). This conceptual notion is empirically substantiated by the findings of Xiao et al. (2020) who indicated a positive relationship between audit effort and audit quality. Specifically, we address the following research questions:

Our approach differs from the extant literature in which researchers evaluate the overall quality of published sustainability assurance statements against indices derived from either a combination of reporting and assurance standards (O'Dwyer and Owen, 2005; Martínez‐Ferrero et al., 2018) or alternatively through a content analysis of published sustainability assurance statements (Zorio et al., 2013; Bollas‐Araya et al., 2019). There are two limitations to this approach. Firstly, “quality” as measured in the literature typically refers to the overall quality of published sustainability assurance statements which is evaluated against scoring indices examining a list of elements that typically fall within four headings:

1. the overall format of published sustainability assurance statements (e.g. provision of information on the title, addressee, name of assuror etc.);

2. the provision of information on the assurance process (including a discussion of the scope and objectives of the engagement, competencies of assuror, evaluation criteria used, assurance standards used, and summary of work performed etc.);

3. whether the assuror provides recommendations or not ^[4]; and

4. the phrasing of the final assurance opinion (e.g. materiality, completeness, responsiveness etc.) offered by the assurance provider (Perego, 2009; Farooq et al., 2023).

While such studies/indices offer useful insights, the scoring methodology used is binary and each item within a category is given a binary score of either 1 or 0. Therefore, if the assurance statement provides any information on how the assurance engagement was undertaken/assurance procedures adopted this is given a score of 1 and if not 0. The limitation of this approach is that it fails to consider what exactly was done by the assuror, i.e. the extent/number of assurance procedures undertaken by the practitioner.

Secondly, this approach runs the risk of biased scoring as professional accounting bodies bar their members from using assurance standards issued by non-accounting standard setters (Farooq and De Villiers, 2019). At the same time non-accountants are free from such restrictions and will often lean towards using assurance standards issued by non-accounting standard setters (Farooq and De Villiers, 2019, 2020). Additionally, the IAASB’s ISAE 3000 is a generic standard (Farooq and De Villiers, 2020) and is not specifically developed for sustainability assurance engagements, while AccountAbility’s AA 1000 assurance standard is a specialist standard but tends not to be used by accountants (Alsahali and Malagueño, 2022; Farooq and De Villiers, 2020). Thus, scoring is predicated on reference to the assurance standards used rather than an in-depth analysis of the exact work carried out by the assurers. This approach offers users limited details about the work performed by assurers (Gillet, 2012).

As the issue of greenwashing of sustainability reports has been on the rise in recent times (Farooq et al., 2023), it has become more important to gain an in-depth understanding of the nature and extent of assurance procedures used by practitioners to achieve their objectives, as well as the factors that influence their choice of assurance procedures for better clarity on the quality and consistency of assurance work performed. Thus, we extend the literature on sustainability assurance by focusing on a direct measure for sustainability assurance procedures used and by highlighting the role of BGD in shaping sustainability assurance in the presence of other important attributes of assurers.

Conceptually, the extent of assurance procedures is a function of client risk (Tarquinio and Rossi, 2017). While assurers favour direct evidence gathering assurance procedures for high-risk clients, they tend to rely on client generated evidence for low-risk clients (Simnett, 2012). Empirically, Du et al. (2023) asserted an increase in assurance procedures for high-risk clients based on control risk and low client integrity. Similarly, assurance procedures used for environmental audits are dependent on client’s control risk (Khalmurzayevna et al., 2023). Consistent with these assertions, we extend these arguments to the context of sustainability assurance to examine the role of BGD in shaping the extent of assurance procedures used by assurers and further highlight the relevance of assurer characteristics in this relationship. For reporting entities, sustainability assurance may either be exploited as a legitimising tool that provides stakeholders with a false sense of confidence in their disclosures or used genuinely to enhance disclosure credibility (Farooq et al., 2023).

BGD, an attribute of corporate governance, influences organisational outcomes (Cumming et al., 2015). Specifically, its relevance has been noted for audit quality (Lai et al., 2017; He et al., 2021; Condie et al., 2023). Chin and Chi (2008) report a positive relationship between female auditors and the level of discretionary accruals. Condie et al. (2023) have also suggested a positive association between audit partner gender diversity and office-level audit quality. These observations from prior studies are indicative of the importance of gender diversity for audit quality. As board gender diversity has been linked to the demand for a higher audit effort (Lai et al., 2017), it has the potential to influence sustainability assurance quality through an insistence on use of a broader range of assurance procedures.

We use managerial capture concept to guide our analysis. The concept of managerial capture, within the context of sustainability assurance, refers to managers using sustainability assurance to further their interests (legitimise sustainability disclosures) while frustrating attempts to promoting disclosure credibility, transparency and accountability (Darus et al., 2014; Farooq and De Villiers, 2020). Sustainability assurance is a voluntary undertaking and in the absence of regulation, managers can restrict assurance practitioners from undertaking extensive procedures. In this way managers can avoid the legal and reputational risks associated with a qualified assurance opinion while enhancing perceptions of credibility amongst unaware/non-expert stakeholders/readers. Thus, a gender diverse board driven by managerial capture (and incentivised by a legitimising orientation to sustainability assurance) may be less interested in demanding a thorough investigation of their disclosures, thereby limiting the extent of sustainability assurance procedures undertaken by the practitioner in the engagement (Smith et al., 2011). Consequently, sustainability assurance tends to be more performative in nature without much substantial commitment from the board (Michelon et al., 2019), making BGD less effective in promoting high quality sustainability assurance engagements. Conversely, a gender diverse board which does not intend to subject the sustainability assurance engagement to managerial capture (and incentivised by a stakeholder orientation to sustainability assurance) will be more interested in demanding for a thorough assurance engagement where a broad range of assurance procedures are brought to bear by assurance practitioners. Thus, BGD may promote the use of extensive assurance procedures.

Furthermore, the asserted relationship between the number of assurance procedures used and BGD is likely to be influenced by the level of assurance (LOA) offered by assurers. Assurance standards, such as ISAE 3000 and AA1000AS require assurers to use more extensive assurance procedures when offering reasonable/high levels of assurance. Consequently, the relationship between the number of assurance procedures used and BGD is likely to be enhanced in engagements which offer a reasonable/high LOA. Conversely, sustainability assurance engagements characterised by a limited/low LOA, are likely to deploy a limited range of assurance procedures in the engagement.

Additionally, we argue that the type of sustainability assurer is likely to have implications for the relationship between the extent of assurance procedures used and BGD. The market for sustainability assurance services is characterised by two competing groups; accounting sustainability assurers (ASAPs) and non-accounting sustainability assurers (NASAPs) (Farooq and De Villiers, 2019). Farooq and De Villiers (2019) suggest that while ASAPs are well-versed in determining and establishing assurance procedures, NASAPs have greater expertise in the subject matter, i.e. sustainability. As these two groups bring different skill sets, they are likely to have differences in their approaches to sustainability assurance work, including the extent of assurance procedures undertaken. The extant literature notes that sustainability assurance quality is higher when the assurance provider is an accounting firm (García‐Sánchez et al., 2022; Perego, 2009; Martínez‐Ferrero et al., 2018). However, these studies do not focus in on the extent of assurance procedures undertaken by practitioners. Thus, assurer type may influence the relationship between the extent of assurance procedures used and BGD.

We focus on multinational companies on the list of Fortune Global 500 and assess the implication of BGD for sustainability assurance procedures. This is against the backdrop of an increase in the rate of sustainability assurance among some members of Fortune Global 500 (KPMG, 2024). Further, the existing literature has predominantly focused on European contexts (Farooq et al., 2023), and thus, a more global representation through the Fortune Global 500 would offer rich insights into the nature of sustainability assurance. Thus, with the uptake in greenwashing of sustainability report in recent time, coupled with an increase in sustainability assurance of these reports, it has become more relevant to articulate the role of BGD in shaping assurance procedures as an effective attribute of corporate governance.

Our study differs from extant literature (Alsahali and Malagueño, 2022; Hazaea et al., 2022) in several ways. Prior studies have examined the quality of sustainability assurance (García‐Sánchez et al., 2019; Kılıç et al., 2021) as well as the drivers for sustainability assurance quality (Hazaea et al., 2022). We extend this paradigm of sustainability assurance literature by considering the sustainability assurance procedures used by assurers and the factors that influence the choice of assurance procedures. Specifically, we highlight the association between sustainability assurance procedures used and BGD and articulate the role of key attributes of sustainability assurance that moderate this relationship. We provide evidence on a more direct measure of the sustainability assurance premised on the actual evidence gathering procedures used, highlighting the diversity in sustainability assurance procedures used. Additionally, we respond to criticism of sustainability assurance as a potential tool for providing credibility for sustainability reports without much substance. This is achieved by assessing the potential theoretical framework that informs management’s pursuit of sustainability assurance in a voluntary setting, particularly in relation to the sustainability assurance procedures employed.

We examine a sample of Global Fortune 500 firms from the period 2016–2021 using firm-year observations of 1148. We use the number of sustainability assurance procedures applied (APs) as the measure for sustainability assurance effort and the proportion of female directors as the measure for BGD. We use an ordinary least squares (OLS) estimation model with fixed effects, the difference-in-difference (DID) research design, and the Two-stage Instrumental Variable method, among others, as our methodological approaches to analyse the relevant data set for this study. Our findings are in line with our predictions and are robust to control variables for other factors of APs. BGD is positively associated with the APs. Furthermore, this relationship is stronger when the assurance work relates to reasonable assurance as against limited assurance. Similarly, when the assurer is an ASAP, the relationship between BGD and APs is stronger as compared to a NASAP. Our results remain consistent under the following model specifications: two-stage instrumental variable model; lead-lag model; and propensity score matched DID model.

Our study contributes to the literature on sustainability assurance in diverse ways. Firstly, we consider sustainability assurance in the context of the assurers’ effort exerted in the conduct of the assurance work using the number of specific assurance procedures applied in the evidence gathering process. This allows us to assess sustainability assurance based on the actual work documented by assurers as being carried out for the assurance work. This provides a direct measure of sustainability assurance quality from a performance-based perspective rather than an indirect measure of sustainability assurance quality. We thus extend the literature on determinants of sustainability assurance quality by highlighting the relevance of corporate governance attribute, BGD, in shaping sustainability assurance quality from a performance-based perspective. Secondly, we highlight the theoretical orientation that is likely to inform a firm’s engagement in sustainability assurance from a performance-based perspective. Our findings provide insight about the preference for a stakeholder orientation to sustainability assurance by a gender diverse board when quality of sustainability assurance is premised on the assurance work done in the gathering of evidence in the absence of managerial capture. Additionally, findings provide stakeholders with an insight into the extent to which sustainability assurance may contribute to shaping sustainability reporting for a better sustainable development agenda based on the availability of high-quality non-financial information.

Furthermore, the extant literature (García‐Sánchez et al., 2019; Kılıç et al., 2021) have indicated the importance of attributes of sustainability assurance in shaping the quality of sustainability assurance. We focus on the type of assurer and the LOA to provide a better contextual narrative for a performance-based understanding of sustainability assurance quality and thus provide stakeholders of sustainability assurance with a better insight on the moderating factors of performance-based sustainability assurance quality. We thus extend the literature on sustainability assurance quality by highlighting intrinsic attributes of sustainability assurance which interact to affect the relationship between sustainability assurance procedures used and BGD. Finally, our findings provide regulators and policymakers with a deeper insight in shaping regulatory frameworks for high-quality sustainability assurance.

The subsequent sections of this paper are structured as follows. The next section of the paper presents an overview of BGD and sustainability assurance. Section 3 focuses on the literature review and hypothesis development with Section 4 highlighting the research method. The results are presented in Section 5. Section 6 concludes the paper.

Assurance procedures used in sustainability assurance engagements include “Inquiry, observation, inspection, computation, confirmation, analytical procedures” (Wallage, 2000, p. 61). However, Wallage (2000) concedes that given the nature of sustainability reporting, assurers will need to apply new and creative assurance techniques to collect evidence to support their opinion. Nonetheless, research shows that the assurance procedures adopted by practitioners, particularly ASAPs, are essentially those applied in traditional financial audits (Farooq and De Villiers, 2019, 2020). This is not surprising given that ASAPs must use ISAE3000.

The two dominant frameworks used by assurers include ISAE3000 and the AA1000AS. While ISAE3000 is very popular among ASAPs, NASAPs tend to prefer AA1000AS (Farooq and De Villiers, 2019, 2020). The key difference between the two is that ISAE3000 is a generic standard developed by the IAASB to assist practitioners in a broad range of engagements where the focus is not on historic financial information (as is the case in a financial audit or financial review). However, the standard reads as a condensed version of ISA’s and essentially exports financial audit methodologies (including audit procedures) to non-financial assurance engagements. In comparison, AA1000AS is designed specifically for sustainability assurance engagements. Importantly, the two standards are often used by NASAPs in combination (Table 1).

Furthermore, sustainability assurance statements often contain vague and ambiguous language (Deegan et al., 2006; Janggu et al., 2013) and provide little information on the assurance procedures undertaken by the assurer. Gillet (2012) indicated the lack of details about work done by assurers.

There is no consensus in the literature on assurer type and the quality of published sustainability assurance statements. O'Dwyer and Owen (2005) has indicated that NASAPs issued assurance statements that contained more depth and breadth of discussion. In comparison, Bollas‐Araya et al. (2019) conclude that ASAPs issue higher quality assurance statements compared to NASAPs. However, these studies tend to use indices constructed from assurance standards (ISAE3000 and AA1000AS) to score assurance statements based on their ability to provide some reference to a list of elements such as addressee, assurer independence, assurance standard used and assurance procedures performed. While this approach offers a useful overview of sustainability assurance practice, it does not provide detailed information on the work performed by the assurer and officially disclosed to users in the sustainability assurance statement.

Moreover, the sustainability assurance process for multinational firms is characterised by a higher degree of variability across countries as well as among firms (Farooq and De Villiers, 2019; Yan et al., 2022). Yan et al. (2022) argued that the lack of a unified framework, the use of assurers with different skill set and expertise, coupled with the stark difference in the sustainability assurance infrastructure for firms, the quality and reliability of sustainability assurance process are likely to differ. As sustainability assurance is primarily undertaken to limit information risk and increase credibility, it remains important to provide an understanding of some of the vital elements that influence sustainability assurance in the context of evidence gathering which ultimately impact the quality and reliability of sustainability assurance.

The issue of BGD among multinational firms has received considerable academic attention with many multinational firms implementing policies to increase BGD in response to regulatory requirements as well as its potential value-adding proposition (Oliveira and Zhang, 2022; World Economic Forum, 2023). The proportion of female directors on US corporate boards increased from 10% to 28% from 2000 to 2020 (Oliveira and Zhang, 2022). Similarly, World Economic Forum (2023) reports an overall increase in the proportion of female as at the first quarter of 2023 of 10.5% with yearly increase of 0.6%. However, the World Economic Forum (2023) extends the antecedents of BGD to include the business case argument for BGD.

These observations from the literature imply two broad categories of reasoning for BGD among multinational firms; satisfying social and regulatory requirements and/or a belief in the potential for BGD to enhance organisational outcomes. While the social and regulatory requirement argument explains the insignificant relationship between BGD and some organisational outcomes for multinational firms, the business case argument explains the positive association between BGD and other organisational outcomes (Oliveira and Zhang, 2022; World Economic Forum, 2023). Under the social and regulatory argument, multinational firms tend to pursue BGD without any substantial commitment to its eventual success. For instance, Chapple and Humphrey (2014) reported the irrelevance of BGD for stock portfolio performance when social pressure and regulatory antecedents are the drivers of gender diversity in firms.

Nonetheless, the business case argument for BGD has received some validation (Alkhawaja et al., 2023; Chen et al., 2019). Alkhawaja et al. (2023) found a positive relationship between BGD and ESG disclosures, while Chen et al. (2019) identified a negative relationship between reputational risk and BGD. Thus, BGD’s role in corporate governance among multinational firms remains uncertain.

The increasing trends for BGD and sustainability assurance respectively among multinational firms have ignited scholarly enquiry into the potential implications of these trends for corporate governance on one hand (Alkhawaja et al., 2023) and sustainability assurance on the other hand (Farooq et al., 2023). Consequently, with the uptake in the greenwashing of sustainability reports (Farooq et al., 2023), the need to examine the quality and credibility of sustainability assurance and the factors that account for these outcomes have gained some prominence among academic scholar (Hazaea et al., 2022). An important question to be addressed in this context relates to the implication of BGD for sustainability assurance in an era of an increase in greenwashing and a rise in the proportion of BGD.

Considering the inherent limitations of the sustainability assurance process (Yan et al., 2022), could BGD serve a parameter for shaping sustainability assurance for a more credible sustainability assurance process? Although prior studies have examined the determinants of sustainability assurance (Martínez‐Ferrero et al., 2018), there has not been a consideration for the implication of BGD for sustainability assurance process in terms of the assurance procedures applied in the evidence gathering process. As BGD has been linked to lower reputational and financial risks, do these benefits of BGD have a spillover effect in shaping sustainability assurance in terms of the sustainability assurance procedures applied in the audit of a sustainability report. We examine the relationship between BGD and sustainability assurance procedures used and further highlight the moderating effects of other attributes of sustainability assurance to help address these concerns. This study offers a deeper insight into a potential disciplinary effect of board gender diversity for sustainability assurance quality in the broader context of corporate governance and sustainability assurance to help emphasise the critical factors of performance-based narratives within the sustainability assurance literature.

We employ managerial capture as the theoretical framework to explain the potential link between BGD and sustainability assurance procedures used (Al-Shaer and Zaman, 2016; Alsahali and Malagueño, 2022; Sánchez-Sancho et al., 2024). Premised on managerial capture, attributes of corporate governance may be limited in their effectiveness to promote positive organisational outcomes (Baker, 2010). Baker (2010) explains managerial capture as the process whereby the agenda and outcome of an activity is subverted by management to attain its interest. In this regard, Tinker and Gray (2003) has described some corporate engagements to be ineffective in promoting corporate accountability. Predicated on this argument, corporate governance mechanisms may foster positive corporate outcomes in the absence of managerial capture (Baker, 2010).

We apply this line of reasoning to the context of BGD as a corporate governance attribute. We argue that BGD has the potential to either enhance or limit an entity’s ability to achieve its intended organisational outcomes. In a situation where BGD is influenced by the ethos of managerial capture, it would have an adverse effect on organisational outcomes. For instance, Owen et al. (2000) submitted that the presence of managerial capture of the verification process for firm disclosures results in a verification method that promote corporate image rather than corporate accountability. In this context, verification processes are shaped by management to reflect their interest and not the concerns of its relevant stakeholders (Ball et al., 2000). Consequently, BGD may adversely affect APs. Consistent with this notion, Arnaboldi et al. (2020) report an overall observation of the non-consequential effect of BGD on bank performance but note the context-specific nature of their findings. Similarly, Chapple and Humphrey (2014) reported an insignificant relationship between BGD and stock portfolio performance. Furthermore, females on audit committee have no relationship with audit quality (Alhababsah and Yekini, 2021). These outcomes indicate the ineffectiveness of BGD in the presence of managerial capture.

Conversely, BGD may be resistant to the influence of managerial capture and have a positive impact on APs. The presence of BGD may account for better risk management and the promotion of better social and environmental goals (Cumming et al., 2015; Eckel and Füllbrunn, 2015). In this context, BGD may promote the use of diverse sustainability assurance procedures to ensure the attainment of a better assurance process for the verification of sustainability reporting. Al-Shaer and Zaman (2016) report a positive relationship between BGD and the quality of sustainability reporting. They suggest that greater attention to social responsibility and stakeholder concern are the main reasons for their findings. Similarly, Erin and Ackers (2024) indicate a significant and positive impact of BGD on the sustainability reporting practices of firm. Chen et al. (2019) posited a negative association between BGD and reputational risk highlighting the cautious nature of a gender diverse board. Furthermore, Alkhawaja et al. (2023) observe a positive association between BGD and ESG disclosure. Additionally, females on boards have a complementary relationship with audit quality (Abdel-Meguid et al., 2023; Cameran et al., 2018). As BGD tend to promote a better sustainability reporting process, there is a greater potential for a spill-over effect of BGD for sustainability assurance.

Based the theoretical and empirical evidence stated above, we argue that BGD and the APs are likely to be positively related. We then formulate out first hypothesis as follows:

Moreover, we assess the implication of the LOA required for the relationship between BGD and the APs. The narrative on LOA in audit posits that, for any given assurance work, the nature and type of assurance procedures used should commensurate with the LOA required for the work to be done (Hasan et al., 2003). Hasan et al. (2003) suggest that while reasonable assurance is associated with a more rigorous assurance process, limited assurance requires a less rigorous assurance process. This variation in the nature and scope of assurance procedures applied in reasonable assurance work relative to limited assurance work is likely to influence the number of assurance procedures used in a sustainability assurance work. Consistent with this line of argument, we expect the association between BGD and APs to be stronger when the LOA required is reasonable relative to limited LOA. In view of this, we formulate the following hypothesis:

Additionally, we investigate whether the type of assurer used for sustainability assurance has an implication for the relationship between BGD and the APs. Farooq and de Villiers (2019) highlighted the substantial differences between ASAPs and NASAPs. Regarding subject matter knowledge, while ASAPs are highly acquainted with financial audit, NASAPs have a better understanding of elements of sustainability assurance. This disparity in knowledge base has resulted in ASAPs focusing on the accuracy and reliability of the content of sustainability reports while NASAPs tend to cater for the completeness and balanced nature of sustainability reports (Farooq and de Villiers, 2019). The resulting difference may bring about a disparity in the scope of sustainability assurance procedures used by these assurers as corroborated by Martínez‐Ferrero et al. (2018).

Furthermore, while ASAPs in the context of the Big4 may have the capability to leverage on economies of scale in their assurance work, many NASAPs are limited by their predominant local and regional reach (Mock et al., 2013). These efficiencies have the potential to shape assurance procedures used by ASAPs differently as compared to NASAPs. As Mock et al. (2013) assert a potential reduction in client fee for ASAP due to economies of scale, ASAPs may modify their assurance procedures to attain maximum efficiencies with minimal clientele fees.

These variations in the knowledge base and skills coupled with the differences in scope and context of operation for ASAPs and NASAPs are likely to shape the sustainability assurance procedures used by ASAPs differently as compared to NASAPs. In line with these submissions, we expect the type of assurer used in an assurance work to moderate the relationship between BGD and APs. We thus formulate the following hypothesis:

We use the list of Fortune Global 500 and focus on the sample period from 2016 to 2021. The use of Fortune Global 500 is premised on the reported dominance of Fortune Global 500 listed firms in the pursuit of sustainability assurance (KPMG, 2024). Additionally, previous studies have used this sample frame as their reference point in determining the relevant sample for sustainability related studies (Crespy and Miller, 2011; KPMG, 2024). Our sample period is influenced by the availability of data on sustainability assurance procedure disclosures in the sustainability reports of firms. Our initial sample is made up of 249 firms with 1494 firm-year observations. We delete 346 firm-year observations due to missing data resulting in a final sample of 249 firms with firm-year observations of 1148.

We source for data in two main ways. For data on sustainability assurance procedures and other related assurance attributes, we resort to the annual sustainability report of our sample firms and manually collect the relevant data points. On the other hand, we use Refinitiv ESG and Worldscope databases to collect data on corporate governance variables and accounting fundamental variables.

Table 2 describes the sample selection process and show the sample distribution by industry. 24.91% of the firm-year observations belong to firms that fall under finance while 15.51% of the firm-year observations belong to firms that fall under wholesale, retail and some services. These industry sectors dominate the distribution of our sample based on the Fama-French 12 (12) industry classification.

We use the number of sustainability assurance procedures used (APs) as highlighted in sustainability reports as our measure for sustainability assurance. APs reflects the number of various assurance processes used in the audit of sustainability report which have been discussed in a firm’s sustainability report by the assurer. The observed score for APs ranges from 1 representing the least score to 13 which represents the maximum score for the number of assurance procedures used. A higher score suggests the use of varied assurance procedures in the gathering of assurance evidence.

The explanatory variable for our study relates to BGD. We use proportion of female directors on corporate boards as the measure for BGD. The observed score for BGD ranges from a minimum value of 0 to a maximum of 62.5%. Furthermore, we use other measures of BGD for sensitivity analysis. We use the number of female directors (NFD) and Blau index (BBGD).

We use OLS regression estimation to test the hypotheses of our study. Regarding H1, we estimate the following model to test the relationship between BGD and APs:

where APs captures the number of sustainability assurance procedures used as noted above and BGD represents board gender diversity as defined previously. Based on H1, we expect α₁ to be positive and significant in equation (1). Controls_i,t are control variables included in our model specification. We control for firm level characteristics and sustainability assurance characteristics that are likely to shape APs.

In line with the extant literature, we include other corporate governance variables in our model (Dutta, 2020). We control for board independence (BIND), board size (BSIZE), CEO duality (CEOD), audit committee expertise (AUDITCEXP) and effectiveness of corporate governance process (GOV). To the extent that these attributes of corporate governance are working to enhance the overall control mechanism of a firm, they should limit firm risk and as such result in a reduction in the APs. Additionally, we control for other firm characteristics such as firm size (FSIZE), return on investment capital (ROIC) and leverage (TDTC). Furthermore, we control for key sustainability assurance characteristics including LOA, type of assurer (TYPEAUD) and the scope of assurance (NOA) (Farooq and de Villiers, 2019).

To test H2 and H3, we expand equation (1) to include the interaction terms BGD×LOA and BGD×TYPEAUD in equations (2) and (3) respectively. Below are specified models for H2 and H3:

From equation (2), BGD_i,t × LOA_i,t represents the interaction term for board gender diversity and LOA. We also include the same control variables as applied in equation (1). We expect α₂ to be positive and significant.

From equation (3), BGD_i,t × TYPEAUD_i,t represents the interaction term for board gender diversity and type of assurer. We include the same control variables as applied in equation (1). We expect β₂ to be positive and significant.

All variable definitions have been provided in the  Appendix.

Table 3 Panel A presents the descriptive statistics for the dependent, the explanatory and the control variables of the study. For APs, we observe the use of at least 6 sustainability assurance procedures on average by assurers, with the dominant being the use of “review of company standards/criteria and its challenges” in the evidence gathering process (see Panel B). For BGD, 23.98% of board members of the firm in our sample are females on average with a standard deviation of 14.52%. This indicates a great disparity in the level of board gender diversity for firms in our sample. This observation is fairly consistent with the findings of Oliveira and Zhang (2022). Additionally, our sample consist of profitable firms with an average return on investment capital of 7.56% and a high level of debt to total capital of 48.59% on the average. From a corporate governance perspective, the firms in our sample have relatively effective corporate governance process with a score of 71.5% characterised by 63.14% of independent directors. In terms of the LOA, most of the assurers issued limited assurance, with most of these coming from ASAPs (Farooq et al., 2023).

In terms of assurer type, both ASAPs and NASAPs appear to deploy a broad range of assurance procedures in line with the provisions of assurance standards (AA1000AS and ISAE3000). However, ASAPs have higher numbers for the deployment of various assurance procedures in each category relative to NASAPs. For instance, while data checks for completeness was used 170 times by ASAPs, NASAPs applied it only 51 times in their engagement works. Similarly, checks for data adequacy was used 407 times by ASAPs but only 269 times by NASAPs. This trend is consistent for the remaining assurance procedures with ASAPs recording higher numbers for usage rate as compared to NASAP. Additionally, and in line with the extant literature (Farooq et al., 2023), our correlation analysis indicates a positive association between the type of assurer and LOA offered. We find that ASAPs’ assurance engagement work requires the provision of a higher number of reasonable assurance (more reasonable assurance as compared to limited assurance) relative to NASAPs. Comparatively, ASAPs use analytical procedures more, while NASAPs prefer review of company standards/criteria and its challenges. Furthermore, among ASAPs, Deloitte has the least amount of assurance procedures applied in its assurance engagement work, while PwC, EY and KPMG deploy a greater breadth of assurance procedures.

Regarding the level of correlation between the independent variables of our model, the highest correlation of −0.624 is observed between FSIZE and BGD followed by 0.449 for AUDITCEXP and BGD (untabulated). We employ both correlation matrix and Variance Inflation Factor (VIF) to assess the potential impact of multicollinearity for our models. We obtained a VIF of 2.49 which is below the threshold of 10. This is indicative of the fact that our results are not subject to problems of multicollinearity.

Table 4 shows the regression results for H1 under column (1). The observed regression coefficient for the relation between BGD and APs is positive and statistically significant at 1% (Coefficient = 0.052). This is equivalent to a 5.2% increase in the APs for a 1% increase in board gender diversity. This outcome is equivalent to 0.36 of the standard deviation of APs for BGD. This implies that the difference is economically meaningful (i.e. 0.052 / 0.145). This observation is consistent with the findings of Cicchiello et al. (2021) where board gender diversity promotes sustainability reporting, the engagement of an external assurance provider and an increase in corporate sustainability practices (Nadeem et al., 2017). This confirms the relevance of board gender diversity in promoting various sustainability related outcomes.

Under column 2, the results for H2 are reported. The coefficient of BGD×LOA is positive and statistically significant at p-value of 1% with a value of 0.056. This result indicates that in the presence of a reasonable LOA, the relationship between BGD and APs is stronger, accounting for the use of a higher number of assurance procedures. As such, LOA act as an enabling moderator for the association between BGD and APs. This observation is consistent with the submissions of Hasan et al. (2003) regarding the role of LOA in assurance work. From an economic significance perspective, one standard deviation change in BGD×LOA is associated with 0.344 (0.056 / 0.162) increase in APs.

Column 3 of Table 4 shows the regression results for H3. The coefficient of BGD×TYPEAUD is positive and statistically significant at a p-value of 1% (coefficient = 0.031). This result shows that the type of assurer used for an assurance work moderates the relationship between BGD and APs. This observation is consistent with the submission of Mock et al. (2013) who highlighted the implication of type of assurer as well as the findings of Harindahyani and Agustia (2023). Our finding is also in line with the observations of Ruiz-Barbadillo and Martínez-Ferrero (2020) who noted the relevance of industry specialisation for sustainability assurance quality.

Overall, it can be said that board gender diversity plays a disciplinary role in the context of managerial capture to enhance sustainability assurance quality with assurer type and LOA strengthen this relationship as complementary moderators. Nonetheless, it should be noted that an increase in the number of assurance procedures used may reflect an ill-designed assurance engagement process. However, controlling for the scope of the assurance work minimises this potential outcome for our stated model. Impliedly, the adverse impact of managerial capture of the assurance process is not evident in this context.

Regarding control variables, board independent is negatively related to the APs. On the other hand, we observe a positive relationship between CEO duality and APs, as CEO duality is deemed to be a risk-increasing attribute of corporate governance. However, board size, audit committee expertise and effectiveness of corporate governance process do not have a significant relationship with APs. Additionally, return on investment capital and leverage are negatively related to APs, while firm size has an insignificant association with APs. LOA, type of assurer and the scope of the assurance are positively related to APs confirming the prior observations from the assurance literature. All our models are highly significant with adjusted R² of 26%.

We conduct series of additional tests that help in addressing any potential endogenous issues associated with our specified models.

First, we employ a two-stage least squares (2SLS) instrumental variable (IV) approach to address the potential issues of simultaneity and omitted variables. We identify an IV which is related to BGD, but it is unrelated to APs. We employ the board gender diversity score of a firm’s industry peer. We first identify industry peers for our sample firms. We then use within sample industry peer board gender diversity score by industry and by year to calculate for our IV. Our use of industry peer board gender diversity is premised on the notion that firms in similar industries are likely to mimic each other in the context of board gender diversity policy while its sustainability assurance process will be dependent predominantly on the specific risk associated with the firm in question. Thus, while BGD should be correlated to industry peer BGD, industry peer BGD is not expected to correlate with APs. Premised on this argument, we implement our 2SLS IV approach using a first stage model which is specified as follows:

where BGD_i,t is the board gender diversity for firm i at time t. BGDPeers is the IV represented by the industry peer board gender diversity score. The control variables used in equation (4) is consistent with those applied in equation (1). Table 5 column 1 presents the results of first-stage model. The instrumental variable (BGDPeers) is positively and significantly related to BGD (Coefficient = 0.044, p-value < 0.01) which is consistent with our expectation of a positive correlation between BGDPeers and BGD.

The second stage model is presented with the associated results in Table 5 under columns 2–4. For the second stage model, we use the predicted values estimated from the first stage model as our explanatory variables represented by BGD^, BGD^×LOA and BGD^×TYPEAUD respectively. All the predicted variables, BGD^ (coefficient = 0.924), BGD^×LOA (coefficient = 0.240) and BGD^× TYPEAUD (coefficient = 0.307), are statistically significant at 1% except for BGD^×LOA which is marginally significant at 10%. Overall, the findings of the 2SLS IV model are consistent with our OLS model specification which implies that the results from our main models are not susceptible to issues of omitted variables and simultaneity.

In our second iteration of robustness test, we employ a lead-lag model to substantiate our evidence on causal relationship between BGD and APs. We re-estimate equations (1) to (3) using BGD_t−1, BGD×LOA_t−1 and BGD×TYPEAUD_t−1 in place of BGD_t, BGD×LOA_t and BGD×TYPEAUD_t respectively. The results for the re-estimated model are presented in Table 6. The regression co-efficient of BGD_t−1, BGD×LOA_t−1 and BGD×TYPEAUD_t−1 are all statistically significant at 1% except for BGD×LOA_t−1 which is statistically significant at 5%. The observed results are consistent with our main findings highlighted in Table 4. This observation indicates that a higher level of BGD in the prior period (t - 1) is associated with a higher APs in the current period (t).

In our third iteration of robustness test, we employ a quasi-experimental approach to substantiate the nature of the causal relationship between BGD and APs. We perform a difference-in-difference analysis for this purpose. In this context, we determine whether the changes in APs for firm-year observations where BGD is greater than and/or equal to the industry mean (treatment firm-year observations) is statistically different from the changes in APs for firm-year observations where BGD is less than the industry mean (control firm-year observations). Our treatment firm-year observations reflect firm-year observations with higher BGD relative to the industry mean. Our control firm-year observations reflect firm-year observations with lower BGD relative to the industry mean. Our control sample is made up of matched firm-year observations. We then code the treatment firm-year observations as 1 with the control firm-year observations coded as 0 to create our treatment variable (TREAT). Furthermore, we establish our pre- and post-period using COVID-19 as the exogenous variable for our difference-in-difference analysis. We reason that the pandemic period of COVID-19 brought about additional elements of risk as firms had to substantially alter their operations and system to deal with the challenges of the pandemic. Thus, we use the period before 2019 as our pre-period and the period after 2019 as our post-period for analysis. We code pre-period as 0 with post-period coded as 1 to create our post variable (POST).

To derive our matched sample of treatment and control firm-year observations, we apply a propensity score matching (PSM) technique using the nearest in neighbour approach with a caliper of 0.002 width. Our PSM technique generated a firm-year observations of 152 each for both the treatment and control firm-year observations. We confirm the validity of the PSM technique with a T-test for difference in mean for the firm characteristics of the treatment and control samples. Panel A of Table 7 presents the results for difference in mean test which indicate no significant difference between the treatment sample and control sample.

In executing the difference-in-difference test, we create an interaction term (TREAT×POST) using our treatment variable and post variable. We use the interaction term to assess whether the change in APs for pre-period is statistically and incrementally different from the change in APs for the post-period between high BGD sample and low BGD sample.

Table 7 Panel B presents the regression result for the difference-in-difference test. The regression coefficient of the interaction term (TREAT×POST) is positive and marginally significant (coefficient = 1.147, p-value < 0.1). This result suggests that the incremental gain in APs is associated with an increase in BGD which is marginally higher for our treatment firm-year observations relative to our control firm-year observations.

We assess the sensitivity of our results to the use of alternative measures. In this regard, we employ NFDs and Blau index (Muniandy et al., 2023). These alternative measures are used in the main model under equation (1) to re-estimate our regression coefficients.

We measure board gender diversity using the NFD and the interaction terms for number of female directors and level of assurance (NFD×LOA) and number of female directors and type of assurers (NFD×TYPEAUD) using the product of NFD and LOA and product of NFD and type of assurer respectively. Table 8 Columns 1–3 present the results for our alternative measures, NFD, NFD×LOA and NFD×TYPEAUD. We observe consistent results for all the alternative measures used. However, the coefficient of NFD×TYPEAUD is only marginally significant.

Similarly, we apply Blau index as an alternative measure to calculate for board gender diversity. We determine our Blau index as a function of $1−∑i=1nPi2$⁠, where P_i represent the proportion of board members in each category for gender and n represent the number of categories. We then generate the relevant alternative measures; board gender diversity based on Blau index (BBGD), interaction term for board gender diversity based on Blau index and level of assurance (BBGD×LOA) and the interaction term for board gender diversity based on Blau index and type of assurer (BBGD× TYPEAUD). Table 8 Columns 4–6 show the results for the use of Blau index. The coefficient of all the alternative measures used under Blau index are positive and statistically significant at 1%. This outcome confirms our findings under our main analysis.

Overall, the results of our sensitivity analysis suggest that our findings are not susceptible to the measurement choice for board gender diversity.

The need to provide credibility for sustainability reports has become an important aspect of sustainability reporting as social and environmental information have become critical in the decision-making process of stakeholders (Free et al., 2024). In this vein, the issue of sustainability assurance has received a lot of attention among both academics and practitioners on how to shape sustainability assurance for effectiveness and efficiency (Farooq et al., 2023). Despite this concern, there are limited research studies that provide insight about the relevant parameters that shape sustainability assurance process for effective evidence gathering process with a focus on the actual assurance procedures used. In this study, we provide some evidence on the role of BGD in shaping the sustainability assurance process and further highlight the moderating role of the level of assurance and the type of assurer used. Premised on the argument of the ability of BGD to enhance the effectiveness of corporate governance, we expect BGD to promote sustainability assurance in terms of the assurance procedures used in the assurance work.

We draw on an international context of Global Fortune 500 firms and find evidence that an increase in BGD is associated with an increase in the APs. Furthermore, level of assurance and type of assurer moderate the relationship between BGD and APs. From theoretical perspective, the findings of our study suggest the non-applicability of the concept of managerial capture to the relationship between board gender diversity and sustainability assurance quality. Impliedly, the opportunistic interests of management do not limit the effectiveness of board gender diversity in its positive impact on sustainability assurance quality limiting the theoretical implication of managerial capture for an adverse outcome for sustainability assurance. This outcome confirms the disciplinary effect of board gender diversity within the context of corporate governance.

We address the issue of endogeneity by applying various robustness tests. We use 2SLS IV approach to address the potential issue of omitted variable and simultaneity. Additionally, we apply a lead-lag model to substantiate the causal relationship between BGD and APs. Furthermore, we also apply difference-in-difference PSM model to further substantiate the main findings of our study. Overall, our findings are not susceptible to estimation biases due to omitted variables or reverse causality.

Although our study applies a rigorous methodology, it still has some limitations. The findings of this study are limited by inherent measurement errors in the construct used for APs. Additionally, since our study uses a sample of Global Fortune 500, our findings may be limited in its application to a context of firms whose characteristics are substantially different from those of Global Fortune 500 firms. Nonetheless, as the need to ensure a better sustainability assurance process has become more relevant, our study provide critical evidence on the role of BGD for sustainability assurance procedures with the resultant effect of informing practitioners, policymakers and other relevant stakeholders. Particularly, policymakers may consider policy revisions in the context of board diversity as a means to achieve positive outcomes in the domain of sustainability assurance for assurance quality. Specifically, regulators of sustainability assurance can consider board gender diversity as a relevant attribute in the design of sustainability assurance standard for high-quality sustainability assurance process. Finally, future research study may consider an experimental research design for performance-based sustainability assurance quality.

The authors would like to acknowledge the financial support received from the United Arab Emirates University, College of Business and Economics Annual Research Program (CARP) 2023 no. UAEU CARP 2023.

Table A1