Multiple global CSR pronouncements and CSR harmonisation in Europe

Economic globalisation has increased stakeholders’ social and environmental awareness and pressure from multiple stakeholder groups emphasises how important it has become for corporates to look beyond just profit maximisation. Such responses are referred to as corporate social responsibility (CSR), which requires that firms recognise their economic, legal, ethical and social responsibilities to address pressures and expectations of stakeholders (Carroll, 1999). Growing attention on firms’ CSR activities has led to increased voluntary CSR reporting and global CSR pronouncements on their progress and results relating to environmental and social activities. In light of the globalisation and increased importance of CSR, one should expect convergence of CSR reporting, especially with the “best practices” in Europe. However, various institutional isomorphic pressures make for an inconclusive conclusion and the question remains:

According to institutional theory, a company operates within the bounds and norms of society (Guler et al., 2002) and CSR reporting can work as a mechanism to address pressures from stakeholders in the company’s environment (Neu et al., 1998). However, societal bounds and norms, as well as stakeholder expectations and pressures, vary between countries and regions. Past literature, such as Matten and Moon (2008) and Neu et al. (1998), emphasise that a country’s historic-cultural background, cultural values, differing institutional norms and varying governmental policies and resources will impact how business ethics and CSR are understood. This promotes strong country-of-origin effects in CSR reporting (Chen and Bouvain, 2009; Einwiller et al., 2016; Fortanier et al., 2011; Mazboudi et al., 2020). Kolk (2010) notes that such effects result in a wide variety of reports with significant differences in length, approach, scope and depth of accountability. If CSR reports of multinational enterprises (MNEs) from different countries are dissimilar in content (not harmonised), then a set of isomorphic pressures could have created a unique reaction reflected in these CSR reports.

Voluntary standardisation of CSR reporting has, therefore, gradually been evolving to address significant differences in approach to such issues. Over time, various organisations have developed many different CSR frameworks, guidelines, standards or formats that firms could voluntarily adopt [²] (Selsky and Parker, 2005). Prior literature, such as Fortanier et al. (2011) and Einwiller et al. (2016), suggest that adhering to the same ground rules in preparing CSR reports improves harmonisation and comparability of these reports.

Recent events related to the formation of the International Sustainability Standards Board (ISSB) and development of the European Financial Reporting Advisory Group (EFRAG) task force further highlight the importance of harmonising CSR reporting. Matuszak and Różańska (2017) note that after the Global Financial Crisis, Europe became the most active region in promoting transparency and disclosure of CSR. Since Directive 2014 / 95/EU, requirements for large companies to disclose non-financial and diversity information have sought to improve the consistency and comparability of non-financial information. Adams (2020) suggests that the EU Commission has an opportunity to lead the world in refocussing corporate CSR efforts in a post-COVID-19 world. At the time of writing, companies that are required to comply with Directive 2014 / 95/EU will have to shift to European Sustainability Reporting Standards (ESRS) in financial year 2024, with the first resulting sustainability statements to be published in 2025 (European Commission, 2023). Given this unique setting, a study of harmonisation effects in the EU is particularly interesting.

Similar to many empirical studies examining CSR reporting in a voluntary setting, this raises inevitable challenges, particularly in terms of self-selection, as firms that expect to gain benefits from using CSR pronouncements are more likely to choose to adopt them. This study, therefore, uses a number of econometric approaches including instrumental variables and lagged analysis. The results are robust when addressing endogeneity. Following Obeng et al. (2021) and Rezaee and Tuo (2019), data about firms’ usage of CSR pronouncements from the Global Reporting Initiatives (GRI) database is collected for the period 2015–2019. This study finds that firms’ use of multiple CSR pronouncements has a positive and harmonising effect on CSR reporting across countries. The model includes several control variables to promote reliable results in seeking to answer the research question. Firm characteristics including year, country and industry fixed effects are also controlled for in the model.

Addressing this research question makes several contributions to the literature. First, the international business literature has mixed results and scarce evidence on harmonisation of firms’ CSR reporting in an international setting (Einwiller et al., 2016). This study contributes to the literature by examining how global CSR pronouncements reduce country-of-origin effects, which in turn enriches our understanding of how firms’ reporting becomes harmonised. There are growing calls in this global economy for greater transparency and accountability of all types of organisations, including corporate decisions about whether and how to adopt CSR and engage in CSR communication through global CSR pronouncements. Research on CSR demonstrates significant cross-national differences in CSR disclosure both in quantity and topics covered. Stolowy and Paugam (2018) find that based on a large sample of US and European firms, firms listed on the S&P 500 Index are less likely to issue a separate CSR report. Kolk and Perego (2010) find that companies located in stakeholder-oriented countries, especially European countries, tend to disclose more information on CSR than those in shareholder-oriented countries such as the USA. The findings strongly support the use of an institutional theory to explain the convergence and the divergence of CSR reporting of European companies.

Second, our findings extend the literature on the harmonisation effects of firms using multiple global CSR pronouncements. Prior studies examining harmonisation effects tend to focus on a single year (Fortanier et al., 2011) and with relatively few countries (Einwiller et al., 2016). This study expands on prior research by examining harmonisation effects using longitudinal data across 16 countries over five years in the EU. Tschopp and Nastanski (2014) suggest that CSR reports have a wider audience than that for financial reports. Consequently, CSR reports have diverse objectives for a wide range of interests and preferences within and beyond the firm. This is reflected in the large number of CSR pronouncements made available over past decades, as each of these may have different audiences or objectives to address. Einwiller et al. (2016) note that if companies give more weight to the demands of certain stakeholders, possibly through solely using a particular CSR pronouncement, then comparability of CSR reporting might deteriorate. Therefore, assessing the effects of firms using multiple global CSR pronouncements with complementarity effect should shed additional light on the literature. This paper adds to the debate on CSR disclosures, specifically its measurement. Most studies measure the level of CSR disclosure through the issuance of a separate CSR report or a manually collected disclosure score and do not address the growing concern about a broader stakeholder as well as complementarity effect of CSR pronouncements (Hummel et al., 2024).

The International Accounting Standards Board (IASB) published a consultation paper on sustainability reporting in 2020. The IASB (2020) suggests that firms are making the decision to use single versus multiple CSR pronouncements. As explained further in the following sections, the current study has the potential to inform and support the work of entities, like the ISSB and EFRAG task force, to address this issue because it examines the harmonisation effects of using multiple CSR pronouncements, which is consistent with recent trends in reporting (KPMG, 2022). Efforts towards harmonisation are emphasised by recent progress, which includes a signed Memorandum of Understanding between the GRI and EFRAG and a partnership between the GRI and ISSB (GRI, 2023a, 2023b).

The remainder of this study is organised as follows. Section 2 provides a literature review of relevant articles explaining country-of-origin effects and how global CSR pronouncements influence CSR reporting. Section 3 describes the sample and research methods. Section 4 provides the main findings, and Section 5 provides the results of additional analysis including tests designed to address endogeneity issues. Section 6 provides a summary and conclusion.

According to Fortanier et al. (2011), historically, there were no agreed-upon terms for the status and contents of non-financial reports. This led to various studies of the different types of voluntary disclosures. These ranged from a narrow focus on environmental disclosures to broader reports that considered the environmental, social and economic dimensions of CSR or so-called triple bottom line reporting (Deegan, 2002; Elkington, 1997). Research across this range shows clear differences in the quantity and type of non-financial information, which reflects a strong country-of-origin effect.

Sethi and Elango (1999) explain country-of-origin effects as being the impact of a country’s physical and human resources, political institutions and culturally based characteristics on the competitive position of a firm from this particular country. Neu et al. (1998) finds country-of-origin effects for CSR reporting arise mainly from institutional pressures. Similarly, Buhr and Freedman (2001) and Kolk (2005) show that the extent of a firm’s susceptibility to such pressures differs between firms in relation to the cultural, political and legal peculiarities of their respective home countries. Meek et al. (1995) suggest that national differences in legal requirements for reporting affect voluntary disclosures because strict requirements, or strong institutional environments, may restrain disclosure innovations and, thus, also voluntary reporting. Holland and Foo (2003) confirm this argument holds in relation to environmental management.

Institutional differences that derive from country-of-origin also affect the influence on a firm of its stakeholders. Pressure from stakeholders is relevant for CSR reporting because Meek et al. (1995) contend that non-financial information addresses issues of social accountability. This is valuable for a wider group of stakeholders than a firm’s shareholders alone. According to stakeholder theory, VA der Laan Smith et al. (2005) contend that institutional differences between countries (particularly differences in corporate governance systems and ownership structure) impact how the role of a firm and its stakeholders is defined in each country. Firms from countries with a stakeholder orientation (Norway/Denmark) have higher levels and quality of CSR disclosures than firms in countries with a shareholder orientation (the USA). Based on these findings, VA der Laan Smith et al. (2005) contend that non-financial reporting is used as a way to address stakeholder pressure, but in distinctive ways for each of these countries. Distinctive ways of addressing stakeholder pressure is further emphasised by Bernow et al. (2019) who reveal that the scope and depth of reporting differs significantly because of the subjective choices companies make about their approaches to sustainability reporting. This discretion includes consideration of which pronouncements to follow, which stakeholders to address and which information to release to the public.

In some cases, local or regional institutional forces necessitate continuous adherence to local demands to maintain local legitimacy (Park and Cave, 2018). According to Gruber and Schlegelmilch (2015), CSR activities act as a compromise between stakeholder demands and business interests in the absence of effective governmental mechanisms to support societal aims. The need for such a compromise diminishes in those countries where governments fill societal gaps. For instance, CSR in Sweden has been embedded within institutional norms and regulations such as codified laws and/or consensual agreements between the government and the private sector. Corporate behaviour relating to CSR is mostly derived from state regulations that are negotiated, so is not voluntary. For instance, air pollution regulations were passed only after consulting various stakeholders (Campbell, 2007). Consequently, firms in Sweden tend not to have much discretion over CSR reporting. Mazboudi et al. (2020) find that firms in Sweden show lower adoption of external CSR policies because of the role and power of various institutional arrangements that drive them to prioritise local demands. Their stakeholders, therefore, makes them assume a more significant role in protecting social welfare. Firms in the private sector in Sweden have reduced incentives to gain legitimacy through adopting external CSR policies because they have smaller marginal legitimacy gains because of the role of institutional arrangements in protecting social welfare. This leads to a lower level of international harmonisation in their CSR reports. In contrast, Mazboudi et al. (2020) note that firms in Brazil, a less developed country than Sweden, have more corporate discretion because the state/public sector has a relatively smaller role in guiding corporate CSR activity. Their weak regulatory environment or absence of infrastructure and supporting institutions motivates Brazilian firms to gain local legitimacy by adopting CSR policies that reflect their commitment to the welfare of their stakeholders.

Besides the effects of local institutional arrangements and internationalisation, the content of reporting could become more similar if there is a common format for the reporting or standards. For instance, in the financial reporting context, Tarca (2004) confirms that International Accounting Standards (IAS) were introduced to improve comparability and transparency of financial statements in countries around the world. Similarly, Tschopp and Nastanski (2014) note that GRI standards support company-to-company comparisons through creation of a common set of topics for CSR reporting. The wave of global CSR pronouncements that continue to emerge, including the three global CSR pronouncements that are the focus of this study, are promoting similar initiatives relating to non-financial reporting. This study focuses on the initiatives that are recorded in the GRI database that are deemed to have complementarity effect (Zinenko et al., 2015): GRI, United Nations Global Compact (UNGC) and the International Organisation for Standardisation’s (ISO) guidance on how to implement and manage non-financial (in particular environmental) commitments within the organisation (ISO 14001). This study will review how these pronouncements affect CSR reporting next.

The literature on firms’ adherence to CSR pronouncements and adoption of organisational practices draws heavily on institutional theory and legitimacy theory. The three types of isomorphic pressure include compliance with legal and institutional requirements (coercive), peer pressure to imitate best practices to remain competitive and legitimate (mimetic) and adhering to industry norms and standards (normative) (DiMaggio and Powell, 1983). According to Ioannou and Serafeim (2015), CSR implementation has been an area of growing interest. Firms perform CSR activity voluntarily for a variety of reasons, mainly to enhance their legitimacy. They use CSR to create bridges with various stakeholders. Deephouse et al. (2017) suggest that legitimacy has been perceived as a survival-enhancing phenomenon that results from conforming to pre-existing regulatory and pragmatic standards, moral values and cultural-cognitive meaning systems. Legitimacy is gained through a degree of fit between a firm and its institutional context. Therefore, the institutional context is considered an important factor in shaping a corporate’s behaviour in its use of CSR activity as a mechanism to support legitimacy.

According to Matten and Moon (2008), the development of global CSR pronouncements seems to have created an additional coercive pull, explicitly operating on a global level under the influence of international governmental or non-governmental organisations. The three global CSR pronouncements are initiatives of such international organisations that tend to have been developed through a process of multi-stakeholder engagement with a wide variety of governmental, non-governmental and corporate actors, in which, interestingly, MNEs seem to have played a pivotal role (Levy et al., 2010). According to Levy and Kolk (2002), global CSR pronouncements make firms less responsive to their country of origin. Levy and Kolk (2002) investigate how global companies respond to global climate change, and they find two kinds of pressure, namely, divergent and convergent pressure. Divergent pressure consists of home country institutional context and individual company histories, while convergent pressure is related to exposure to a common global industry and the overarching nature of the climate change issue. Levy and Kolk (2002) also find that convergent pressures at this level tend to predominate as the issue matures because companies within the global industry develop a more sophisticated understanding regarding the climate change issue. Once common action is created, other companies’ strategies within the global industry are likely to be similar. Fortanier et al. (2011) contend that global CSR pronouncements create this convergent pressure for CSR reporting and firms become less responsive to their country of origin because they conform to these pronouncements.

On the other hand, global CSR pronouncements might not create enough convergent pressure for CSR reporting. Tschopp and Nastanski (2014) note that less harmonisation of CSR reporting and CSR standards than that of financial accounting should be expected. Firms differ in their CSR practices, reflecting their exposure to various social and economic systems. Companies face diverse stakeholders that create various cultural and institutional pressures. The substantive difference in national institutional arrangements makes CSR “unlikely to be easily transformed into uniform standardized practice across the globe” (Jamali and Neville, 2011, p. 617). In the same vain, Chen and Bouvain (2009) find that the UNGC membership has a limited effect on CSR reporting of leading companies in the USA, the UK, Australia and Germany. Those companies face significant national differences derived from diversity of institutional arrangements that prevent harmonisation of CSR reporting. Ervits (2021) finds symbolic harmonisation of CSR reporting worldwide in China and Western companies (European and the USA). Similarly, Vollero et al. (2022) find that CSR reporting content differs significantly among firms based in different regions (Asia, Europe and North America) except on topics related to CSR identity, which are messages that show the institutional attitude of organisations towards social responsibility. Furthermore, Jeriji et al. (2023) find that investors may consider that the adoption of GRI standards could be misleading, as they penalise companies who adopt GRI standards through a decline in Tobin’s Q. Levy and Kolk (2002) suggest that isomorphic pulls can be very diffuse because the organisational field is not a homogeneous construct. In addition, Westney (1993) contends that the boundary of a field is often difficult to establish. Furthermore, Westney notes that their activities cross industries and national borders, because some firms, especially MNEs, belong to multiple fields. Therefore, Westney argues that a MNE is subject to sometimes incompatible or inconsistent isomorphic pulls because it tries to conform to a wide variety of geographically disparate institutional pressures. Sometimes the powers of convergence are not enough to counter the powers of national institutions. Gooderham et al. (1999) also contend that variations in institutional structures are more likely to generate divergent management practices, which limits convergence efforts. Previous literature (Brammer et al., 2012; Campbell, 2007; DiMaggio and Powell, 1983) shows that dominant local institutions contribute to sustaining implicit, taken-for-granted norms and practices. This discourages harmonisation of CSR reporting.

According to Elkington (1997), the influence of globalisation and growing powers of MNE’s contribute to the diffusion of certain practices across the globe. McGaughey and Cieri (1999) point out convergence is further promoted by developments in communication channels and growth in inter-organisational interdependency and collaboration across countries. Matten and Moon (2008) suggest that institutional isomorphism promotes diffusion because of various forces, which causes firms to converge their CSR reporting. According to Husted et al. (2016), whether the power of convergence supersedes divergence has been an area of recent research attention. Firms may seek global legitimacy rather than mere local legitimacy to address forces of globalisation. Scherer and Palazzo (2011) find an increase in firms adopting strategies to gain global rather than local legitimacy. Fortanier et al. (2011) find that global standards lead to harmonisation of CSR reporting across national contexts. The changes in CSR approaches are attributed to forces related to corporate motivations to earn more legitimacy in the global arena (Lopez and Fornes, 2015). These forces support the harmonisation of CSR reporting. Dowell et al. (2000) point out that the competitive effect of globalisation encourages firms to adhere to severe environmental standards. In doing so, firms could avoid potential clean-up costs and create competitive advantages compared to local firms that merely comply with (less stringent) local standards. The forces also encourage firms to adopt “best practice” at a global level, for example, following each other in joining international coalitions for socially responsible behaviour. Furthermore, normative pressures that are derived from professional associations and educational institutions push firms in the direction of greater adoption of CSR policies, which improves harmonisation of CSR reports. Global reporting pronouncements, such as the GRI and the UNGC, are likely to play an increasing role in shaping CSR reporting adoption at a global level. Russo-Spena et al. (2018) find the harmonisation of CSR reporting content in the global automobile industry through normative isomorphic pressures. The reporting companies use global CSR pronouncements as they are pressured to speak the same CSR language in the global market to remain competitive. They may use these global CSR pronouncements because it is convenient, as they provide a template flexible enough to adapt various isomorphic pressures and fitting enough for their corporate reality (Fortanier et al., 2011). Similarly, Matten and Moon (2020) find a move towards greater harmonisation of CSR reporting because of the increasing standardisation of CSR disclosure through the GRI standards and UNGC. They also note that harmonisation of CSR reporting can be encouraged by mimicking and normative pressures through educational and professional authorities.

Previous studies attempt to examine the effect of global CSR reporting pronouncements on harmonisation, but overall, results are inconclusive. Fortanier et al. (2011) conduct content analysis of non-financial reports of Fortune Global 250 firms for the year 2004 by measuring five binary items indicating if a firm reported on a particular issue in its CSR report. They find harmonisation effects in reporting for those companies that adhere to global CSR pronouncements (GRI G2 and UNGC, among others) concerning issues on the social and environmental dimension of CSR reporting. On the other hand, Chen and Bouvain (2009) find similarities in the content of CSR reports as a result of companies’ membership in the UNGC only in the areas of environment and employees. Their samples are based on companies headquartered in the USA, the UK, Australia and Germany for the year 2006. Seele (2015) also conducts content analysis with a focus on GRI reports for 2012 that hold the highest reporting level (G3.1, A+), but finds significant differences regarding the frequency with which GRI specified indicators are reported. Furthermore, Sherman and DiGuilio (2010) compare two companies each for four different industries (automotive, financial services, pharmaceuticals and sporting goods). They compare the reports for two discrete periods (2006–2007 and 2008–2009) across the USA, Germany and the UK. Their findings are inconclusive as to whether GRI reports become more comparable because of a lack of common measures for reporting. According to Einwiller et al. (2016), inconclusive findings might also be because of different samples, years of reporting and different methods applied. For instance, Fortanier et al. (2011) deductively align their variables with the triple bottom line using manual coding to check CSR report content, while Chen and Bouvain (2009) derive the variables inductively using the software Leximancer. Seele (2015) base their analysis on the coding of indicators specified in the GRI guidelines. In addition, their analysis focuses on a comparison within GRI G3.1 reports with application-level A+; it was not their objective to test whether adherence to the standard leads to relatively more comparability than not adhering to it. The same holds for a study by Sherman and DiGuilio (2010).

Relying on a single pronouncement might impede harmonisation. De George et al. (2016), Jarolim and Öppinger (2012) and Sunder (2011) suggest that adoption of IFRS as a single set of high-quality, principles-based standards eliminates the opportunity to compare alternative practices and learn from them. Sunder (2011) also points out that a single pronouncement set is less likely to cater for local variations in legal, auditing, regulatory and governance regarding reporting across the globe. CSR reporting pronouncement sets tend to target different objectives and users. Similarly, a single pronouncement might reduce harmonisation of CSR reporting because of such limitations.

Zinenko et al. (2015) expand on this, finding that CSR pronouncements should not be treated as separate alternatives, but rather as complementary to each other because of their different objectives and features. The CSR pronouncements could be divided into three categories: codes of conduct; CSR/sustainability reports; and management systems and certification schemes. Codes of conduct are CSR pronouncements, such as UNGC, which are based on widely shared principles and codes of practice applicable to all sectors and organisations. They may be seen as a good starting point for implementing CSR practices (Rasche and Esser, 2006). The UNGC’s main aim is to guide policy to meet fundamental responsibilities in the areas of human rights, labour, environment and anti-corruption (Kell, 2005).

The UNGC has been criticised for being too lenient (Gjølberg, 2009), indicating only symbolic commitment (Perez-Batres et al., 2012) and offering weak compliance mechanisms (Gjølberg, 2011). Still, it can be seen as a force influencing the design, assessment and practice of reporting on CSR (Schneider and Meins, 2012). To address such weak compliance mechanisms, the UNGC does require “communication on progress” (COP) reports from reporting entities every year. Failure to comply is penalised by noting this on the UNGC website. Amer (2018) finds that such entities are penalised in the financial markets (lower average cumulative abnormal return), which implies that stakeholders may be able to pressure UNGC participants to comply with UNGC requirements (normative pressure).

Pronouncements relating to CSR/sustainability reports provide more visibility and accountability than the codes of conduct because they reflect a company’s real public commitment to CSR activities (Guidry and Patten, 2010). The GRI was initially established to become a common sustainability reporting standard and to satisfy demand for companies to report CSR information (Kolk, 2008). The GRI features a multi-stakeholder process and also provides a series of supplements to address sector-specific circumstances and challenges (GRI, 2024).

However, the GRI face criticism by scholars for being overly focused on the reporting process rather than improving social and environmental performance outcomes (Levy et al., 2010). Levy et al. (2010) suggest that the GRI has proved to be more successful in gaining corporate acceptance than in contributing to sustainable development. Arguably, use of the UNGC could complement and compensate for such shortcomings.

ISO standards could further complement the aforementioned pronouncements by providing more practical and manageable steps in implementing CSR activities. ISO is one of the most internationally well-known, standard setting organisations and supports organisations dealing with the ambiguities resulting from the various pressures from stakeholders on private enterprises (Hahn, 2012). Two ISO series that are commonly used for CSR reporting are ISO 26000 and ISO 14000. The goal of ISO 26000 is to offer guidelines to those organisations who consider that respect for society and environment is a critical success factor. The goal of the ISO 14000 series is to provide practical tools for organisations to manage their environmental responsibilities. Mion and Loza Adaui (2019) note that both provide management process schemes, which promote continuous improvement such as Plan-Do-Check-Act. However, one critical concern raised by scholars is that ISO only standardises processes and definitions, rather than focusing on the results and reporting of CSR performance (Schwartz and Tilling, 2009).

These three CSR pronouncements (UNGC, GRI and ISO) have the same overall objective in common: to support and encourage all types of organisations to act in a socially and environmentally responsible way (Rasche, 2009). However, they do this in different ways, which complement each other. Prior literature, such as García-Sánchez et al. (2019) and Zinenko et al. (2015), suggests that the complementary nature of such CSR pronouncements and use of multiple CSR pronouncements increases the quality of the information environment. This may encourage a greater level of harmonisation of CSR reporting as well.

According to Einwiller et al. (2016), despite these inconsistencies in previous findings, there seems to be at least some evidence that adherence to global CSR pronouncements promotes harmonisation. Coombs and Holladay (2013) suggest companies have significant discretion in their CSR reporting given limited requirements. Consequently, companies could choose to define CSR as various activities, from simple philanthropy to employee volunteering to more complex issues such as evaluating environmental impacts and their supply chain. Corporate reporting may focus on favourable topics while intentionally omitting areas where they are much less effective or even operating in irresponsible ways (Seele, 2015). Such obfuscation could become more apparent if most companies report on the same range of topics with common reporting standards. Using similar CSR pronouncements and adhering to the same ground rules in preparing the reports should encourage greater comparability and prevent the problems raised previously (Einwiller et al., 2016; Fortanier et al., 2011; Sherman and DiGuilio, 2010). Einwiller et al. (2016) also point out that harmonisation through reporting standards might reduce biased selective reporting. This is similar to GRI’s aim to encourage transparency and to enhance comparability over time and across companies by creating a global common language for organisations with its extensive reporting framework (GRI, 2020). Therefore, we expect that the adoption of global pronouncements will moderate the country-of-origin effect. For firms that adhere to more of the above-mentioned pronouncements, we also expect domestic institutions to play a less important role in explaining variation of CSR disclosures. This leads to the following hypothesis:

This section provides empirical evidence relating to the effects of CSR pronouncements on harmonisation by using longitudinal data in an international setting. This study draws from the following databases: GRI database, Compustat Global, Bloomberg, Worldscope, DataStream, Eikon and World bank. The sample consists of all EU-firm-years recorded in the GRI database. The year 2015 is selected as the first year of data collection because the European Parliament and Council issued Directive 2014 / 95/EU, which imposed mandatory disclosure of non-financial environmental and social information for public interest entities with more than 500 employees (European Union, 2014). Directive 2014 / 95/EU was enforced on December 2014, so 2015 is the first-year firms’ disclosures are expected to be influenced. The year 2019 is the final year of data collection to avoid the influence of COVID-19 because this is an unusual global event, which could significantly disrupt normal patterns of behaviour. Following Rezaee and Tuo (2019), the study identifies voluntary adopters of global CSR pronouncements from the GRI database. The GRI database also tracks other sustainability reporting information, which is used to ascertain what other CSR pronouncements firms may use in addition to the GRI standards/guidelines. A key advantage of using the GRI database to collect data on global CSR pronouncements is that it includes numerous types of global CSR pronouncements, including GRI, UNGC and ISO.

Firm-level financial data is obtained from Compustat Global and Bloomberg, non-financial data from Refinitiv Worldscope (previously Thomson Reuters ASSET4), DataStream and Eikon, and country-level data is obtained from the World Bank. A total of 2,443 firm-year observations from 636 unique firms from 16 countries are collected. Table 1 details the sample selection process for sample firms.

CSR could be understood in relation to how firms address needs of stakeholders through the adoption of CSR policies (Sánchez and Benito-Hernández, 2015). Following Mazboudi et al. (2020) and Fortanier et al. (2011), this study examines various CSR policies addressing needs identified as being important for a range of key stakeholders, such as community, employees and consumers, to measure the level of harmonisation of CSR reporting. Sánchez and Benito-Hernández (2015) note that those stakeholders are most relevant to understand companies’ CSR. For example, community policies imply a firm’s commitment towards maintaining the firm’s reputation within the general community by being a good citizen. Employment quality policies reflect a firm’s explicit commitment towards, for instance, ensuring good employee relations within the supply chain and maintaining long-term employment growth and stability. Product responsibility policies reflect a firm’s explicit commitment to protecting consumer health and safety by ensuring quality of products and services. This is consistent with the societal role of business towards the community, avoiding improper treatment of employees and safeguarding consumers from faulty production (Harrison and Freeman, 1999) [³]. The Eikon database covers public firms listed in major stock indices and collects CSR data from typical sources such as stock exchange filings, sustainability reports and news sources. Every answer to every data point question proceeds through a multi-step verification and process control, which includes a series of data entry checks, automated quality rules and historical comparisons to keep a high level of accuracy, timelines and quality (Cheng et al., 2014).

This study measures the level of CSR harmonisation (CSR_POLICY) as the average score for ten questions concerning CSR policy adoption. Our dependent variable is developed from Mazboudi et al.’s (2020) method that covers a comprehensive topic for analysing the harmonisation of CSR reporting, and this study adjusts the list of questions to answer the research question.

The ten questions include various social issues that are related to employment quality, health and safety, training and development, diversity, human rights and product responsibility. The list of all questions is available in  Appendix. The answer to each question is equal to 1 if the firm has a related policy and 0 otherwise. To get an average value between 0 and 1, the answers of all questions are then summed and divided by ten. Thus, CSR_POLICY is an ordinal discrete variable that ranges from low CSR policy adoption (value = 0) to high CSR policy adoption (value = 1). A higher value reflects greater similarity of CSR reporting content and harmonisation in CSR reporting. For instance, CSR_POLICY takes the value of 0.1 (1 / 10) if a company adopts only one policy out of the ten policies identified as being important.

Following Ioannou and Serafeim (2012), this study creates a dummy variable for each type of CSR pronouncement used. It equals 1 when a firm is using a particular CSR pronouncement and 0 otherwise. Data is collected from the GRI Sustainability Disclosure Database [⁴]. Prior CSR literature using international samples occasionally uses aggregated CSR measures to study a holistic dimension of CSR. Waddock and Graves (1997, p. 304) suggest the “need for a multidimensional measure applied across a wide range of industries and larger samples of companies”. Therefore, this study uses a total of three global CSR pronouncements. The variable FRAMEWORK is calculated as the average of these, and it is an ordinal discrete variable that ranges from low FRAMEWORK (value = 0) to high FRAMEWORK (value = 1). For instance, FRAMEWORK takes the value of 0.3333 (1 / 3) if a company adopts only one out of the three CSR pronouncements identified as being important. A higher value represents higher adoption of CSR pronouncements.

This study follows prior studies (Fortanier et al., 2011; Mazboudi et al., 2020) and uses the ordinary least squares (OLS) regression model to test the hypothesis. To test the hypothesis, the following model is used:

where CSR_POLICY represents the average of ten CSR policies as described above and FRAMEWORK is the average score for three types of CSR pronouncements. This study includes several control variables to study the effect of using multiple CSR pronouncements on harmonisation. Following Mazboudi et al. (2020), this study measures the degree of internationalisation (FOREIGN_SALES) because they find higher harmonisation of CSR policy adoption as firms internationalise because firms are motivated to gain global legitimacy and keep competitive in the global market. Furthermore, following Mazboudi et al. (2020), Fortanier et al. (2011) and Cai et al. (2012), this study controls for a number of other firm characteristics, namely, leverage (LEV), profitability (PROFIT), firm size (SIZE) and growth opportunities (MTB). LEV is expected to be positively associated with harmonisation because Mazboudi et al. (2020) find that debt enhances the harmonisation of CSR policy. Fortanier et al. (2011) and Mazboudi et al. (2020) suggest that higher PROFIT contributes to higher levels of harmonisation and expect a positive relationship between PROFIT and CSR_POLICY. Following Fortanier et al. (2011), SIZE is expected to be positively associated with harmonised CSR reports because bigger firms become more visible and have more impact on society and environment. Increased public pressure because of these factors leads to increased CSR disclosure. In line with Mazboudi et al.'s (2020) findings, this study expects MTB is positively associated with CSR_POLICY, as firms would be more motivated to adopt CSR_POLICY to gain further legitimacy. This study controls for the number of analysts (ANALYSTS) because Cai et al. (2012) suggest that firms with higher analyst followings gain more public attention and receive greater scrutiny, hence are more likely to engage in CSR.

This study also controls corporate governance variables that might be associated with CSR reporting. According to Appuhami and Tashakor (2017), audit committee independence (AC_INDEPENDENCE) is positively associated with CSR disclosure. Director independence is an important characteristic of audit committees in the monitoring of managers’ actions and improving the level of financial and non-financial disclosures. Dobbs and Van Staden (2016) suggest that shareholder rights (SH_RIGHT) are an important factor in firms’ decisions to prepare CSR reporting. Firms that care about shareholder rights (e.g. protections for minority shareholders) are likely to be encouraged to publish CSR reporting. Existence of a CSR committee (CSR_COMMITTEE) supports enhanced firm CSR performance and reporting by aligning corporate objectives and improving communication with internal and external stakeholders (Kuzey et al., 2021). Arena et al. (2018) find that CSR awards (AWARDS) are associated with higher disclosure. Katmon et al. (2019) find that board diversity (DIVERSITY) affects the level of CSR disclosure.

Consistent with Kendo and Tchakounte (2022), Mazboudi et al. (2020) and Mollagholamali and Rao (2022), following country-level variables affects CSR reporting, namely, institutional environment (INSTITUTION), economic development GDP and legal tradition (CIVIL). Following Lu and Wang (2021), this study measures the degree of institutional environment (INSTITUTION) using the average of six dimensions of Worldwide Governance Indicators: voice and accountability, political stability and absence of violence/terrorism, government effectiveness, regulatory quality, rule of law and control of corruption. A higher value indicates that a country has a stronger institutional environment. GDP is the natural log of GDP per capita. Kendo and Tchakounte (2022) find that GDP is a measure of the financial openness of an economy and represents financial integration and, thus, is associated with reporting quality. La Porta et al. (2008) contend that countries with a civil law framework have a more stakeholder-oriented approach than those which rely on common law. A civil law framework approach is likely to enforce legal requirements for firms regarding CSR. The objective of such an approach is that the law protects the interests of various stakeholders such as employees, consumers and the wider community. La Porta et al. (2008) suggest that CSR is more likely to be a mandatory regulatory requirement rather than a voluntary action in civil law countries. Similarly, Jo et al. (2016) find that firms in civil law countries have higher quality of CSR reporting than in common law countries.

To control unspecified, invariant effects related to years, industries and countries, this study also includes year, industry and country fixed effects in all the models. All variables are defined in  Appendix. β₁ is expected to be positive and significant. It represents the contribution towards harmonisation of CSR reports by adopting CSR policy through using multiple global CSR pronouncements.

Table 2, Panel A displays the descriptive statistics of the variables used in the regression analysis. This study winsorises all variables, except dummy variables, at the 1st and the 99th percentiles to avoid the influence of extreme values. The descriptive statistics reveal that CSR_POLICY has a mean of 0.7649, median of 0.8000, minimum of 0.3000 and maximum of 1.000. These indicate that sample firms satisfy at least three CSR policy focus areas, with most firms satisfying around eight such areas. This value is relatively lower than what Mazboudi et al. (2020) find, and this might potentially be because of the different settings (EU vs Sweden and Brazil). The FRAMEWORK statistics show that sample firms use up to three CSR pronouncements, but the mean (0.6555) and the median (0.6667) suggest that most firms use around two.

The mean of FOREIGN_SALES (52.5860), median (59.21) and maximum (98.71) all suggest that foreign sales indeed have significant potential to affect the harmonisation of most firms. The mean value is relatively higher than what Mazboudi et al. (2020) and Fortanier et al. (2011) find. The LEV mean value indicates that, on average, sample firms have 18.28% debt in their capital structure. This value is close to the 19% found by Li et al. (2016). The mean of PROFIT indicates that 4.62% of total assets are net income, which is consistent with Mazboudi et al.'s (2020) finding of 5.40%. The mean of SIZE is 8.0704, somewhat higher than Rashid's (2016) results. The mean and median of MTB are 2.6876 and 1.7911. Both statistics indicate that sample firms’ stocks are traded at prices above their book values, consistent with what Michaels and Grüning (2017) find (2.15). The mean and median of ANALYSTS are 2.6431 and 2.7726, respectively. This value is relatively higher than what Jo and Harjoto (2014) find (1.771). The mean and the median of AC_INDEPENDENCE are 94.4069% and 100%, consistent with Appuhami and Tashakor (2017), who contend that audit committees should consist only of non-executive directors. The mean value of SH_RIGHT (0.9415) indicates that 94.15% of sample firms use shareholder rights policies. Similarly, the mean value of CSR_COMMITTEE (0.7806) indicates that 78.06% of sample firms have established a CSR committee. This is higher than what Kuzey et al. (2021) find, possibly because of different sample selections (they focussed on the tourism, health care and financial sectors). The mean value of AWARDS (0.4826) indicates that 48.26% of sample firms received awards on their CSR reports. The mean value of DIVERSITY is 26.5788%. The median of INSTITUTION is 1.321. A higher index value represents stronger institutional influence. The mean of the natural logarithm of GDP is US$10.6019, which means that, on average, the sample countries have US$40,200 GDP per capita. This value is consistent with what Yu et al. (2018) find, which is US$37,700 (natural logarithm of US$10.5370). Finally, on average, 72.78% of sample firms are based in civil-law countries.

Panel B, Table 2 provides sample distribution by year. Year-wise distribution shows that the year 2019 accounts for the largest number of observations (582 = 22.82%), followed by 2018 (564 = 23.09%), while 2015 has the lowest number of observations (421 = 17.23%). The low number of observations in 2015 and increasing observation towards 2019 is consistent with KPMG’s (2022) report that shows companies are increasingly publishing CSR reports and using global CSR pronouncements. Our main dependent variable, CSR_POLICY, tends to increase over time, which implies firms’ CSR reporting is gradually harmonising. CSR_POLICY in the year 2015 has the lowest value across the sample period, while the year 2019 has the highest value. FRAMEWORK tends to have similar value over the sample period.

Panel C, Table 2 shows country distributions of sample firms. The sample consists of 16 countries and is dominated by firms from the UK (40.56%), followed by France (10.85%), while Hungary has the lowest number of observations (0.37%). Country-wise descriptive stats reveal that Poland has the lowest CSR_POLICY (0.6397), while Spain has the highest CSR_POLICY (0.8632). The countries that have a high mean value of FRAMEWORK tend to have higher CSR_POLICY, such as Finland, Germany, Portugal and Spain, with Spain the highest FRAMEWORK (0.8962). The exception to this is the UK, which has relatively low FRAMEWORK (0.5254) while having relatively higher CSR_POLICY.

Panel D, Table 2 provides sample distribution by industry. The sample includes companies in a variety of industries classified using the Standard Industry Classification Codes. The industry distribution shows that the sample is dominated by firms from the manufacturing industry (42.12%), followed by the transportation and communications industry (13.63%), while the agriculture, forestry and fishing industry has the lowest number of observations (0.16%). The industry-wise descriptive statistics show that industries that have higher FRAMEWORK tend to have higher CSR_POLICY, similar to the year- and country-wise descriptive statistics. An exception to this pattern is the agriculture, forestry and fishing industry with the highest mean CSR_POLICY and fourth lowest mean for FRAMEWORK. Regarding the mean of FRAMEWORK, the other industry has the highest mean value across the industry, which is 0.7821.

Table 3 provides the correlation matrix of the variables used in the regression models. The correlation matrix shows that firms with higher usage of global CSR pronouncements (FRAMEWORK) are significantly and positively associated with CSR_POLICY, which is consistent with the hypothesis. This implies that firms using global CSR pronouncements tend to have more harmonised CSR reports. Using global CSR pronouncements supports harmonisation of CSR reports by reducing selective reporting (Einwiller et al., 2016), supporting the user to build a reporting system (Hąbek and Wolniak, 2016) that potentially strengthens the normative isomorphism and promotes convergence mechanisms of CSR disclosure (Mazboudi et al., 2020). Consistent with prior literature (Arena et al., 2018; Cai et al., 2012; Mazboudi et al., 2020), SIZE, ANALYSTS and AWARDS are significantly and positively associated with CSR_POLICY, while INSTITUTION is statistically and negatively associated with CSR_POLICY. Further, the correlation matrix shows that the potential for multicollinearity in the regression models is unlikely. Gujarati and Porter (2009) suggest that correlations between variables with values below 0.80 do not create any multicollinearity issues in regression models. Therefore, there is no observable multicollinearity issue in the regression models.

This study applies OLS to examine the relationship between FRAMEWORK and CSR_POLICY. In Table 4, the coefficient of FRAMEWORK is positive and statistically significant (Coefficient = 0.1077 and p < 0.001). The adjusted R² of the model is 49.62%. This suggests that use of multiple global CSR pronouncements is associated with better harmonisation through greater convergence mechanisms of CSR reporting. The greater convergence mechanisms could be enhanced by the complementarity nature of CSR pronouncements (García-Sánchez et al., 2019; Zinenko et al., 2015). Therefore, the results support the hypothesis. In terms of economic significance, on average, a one standard deviation (0.3410) increase in FRAMEWORK is associated with a 3.6726% increase in harmonisation.

Among the control variables, FOREIGN_SALES, LEVERAGE, SIZE, ANALYSTS, CSR_COMMITTEE and AWARDS are positively and significantly related to CSR_POLICY. This suggests that firms with greater internationalisation, leverage, size, analysts following, existence of CSR committee and receiving awards on CSR reporting are positively associated with harmonisation.

CSR_POLICY could be correlated period-to-period. To address the issue of reverse causality, this study uses lagged analysis by using one-year lagged variables. Table 5 provides results for a lagged analysis. Consistent with the main regression analysis, CSR_POLICY has a significant positive association with FRAMEWORK_LAG (Coefficient = 0.0951 and p < 0.001). The results are robust when we examine the effect of one-year lagged variables. The main findings, therefore, still hold after controlling for lagged variables.

Overall, this provides evidence consistent with the hypothesis: use of multiple CSR pronouncements is positively associated with CSR_POLICY. This suggests that use of multiple global CSR pronouncements contributes to reduced country-of-origin differences and, thus, promotes harmonisation of CSR reporting.

Table 6 shows results for the regression analysis excluding observations from countries or industries that hold less than 1% of total observations. This is in line with prior literature such as Tomas and Wang (2019) and Nguyen et al. (2019). Dropping observations from Hungary (nine observations) and the agriculture, forestry and fishing industry (four observations) still provides results consistent with the main regression. CSR_POLICY has a significant positive association with FRAMEWORK (Coefficient = 0.1075 and p < 0.001).

Table 7 provides results for the regression analysis excluding the financial industry. Prior accounting disclosure research often excludes the financial industry because of unique regulatory requirements and operating environments associated with this sector (Leung and Horwitz, 2004). The result is still consistent with the main results after removing the financial industry. CSR _POLICY is positively and significantly associated with FRAMEWORK (Coefficient = 0.1161 and p < 0.001).

In summary, CSR_POLICY remains positively and significantly associated with FRAMEWORK, after excluding samples from small observations and the financial industry.

The major empirical challenge this study faces is the endogenous relationship between voluntary adoption of CSR policy (harmonisation of CSR reports) and CSR pronouncements. Specifically, omitted variables and reverse causality could violate the validity of the results. Consequently, to address endogeneity concerns, this study conducts a series of additional tests.

To address possible endogeneity arising from unobserved, unit-specific confounders in the model, this study performs regression analysis with firm fixed effects. According to Imai and Kim (2019), past literature also uses unit fixed effects regression models for estimating causal effects with longitudinal data. Table 8 shows results for the regression analysis with firm fixed effects and the main result still holds. FRAMEWORK has significant positive coefficients with CSR_POLICY (Coefficient = 0.1077 and p < 0.001).

This study uses a two-stage least squares instrumental variable (IV) approach to address other sources of endogeneity such as omitted variables and simultaneity. IV that is related to adopting CSR practice but has not been identified as a determinant of harmonisation is needed. In line with Sun et al. (2020), this study uses country-industry mean FRAMEWORK (INDCOUN_FRAMEWORK) and country-year mean FRAMEWORK (YEARCOUN_FRAMEWORK) as instruments. These variables are the mean score of FRAMEWORK by country and year and the mean score of FRAMEWORK by country and industry. These instruments represent the average FRAMEWORK score of firms within the same industry, and across years, in a given country [⁵]. This variable is useful because peer effect is likely to shape a firm’s CSR practice and, thus, would be correlated with CSR pronouncements. However, the peer CSR score should not be correlated with the CSR_POLICY of a particular firm, except through the global CSR pronouncements channel. INDCOUN_FRAMEWORK and YEARCOUN_FRAMEWORK are expected to be positively and significantly associated with FRAMEWORK. This study conducts the analysis using the first stage of the two-stage least squares model as follows:

Table 9 shows the results of the IV model. In Column 1, the first stage model shows that INDCOUN_FRAMEWORK and YEARCOUN_FRAMEWORK are positively and significantly associated with FRAMEWORK (coefficient = 0.6633 and p < 0.001; coefficient = 0.7528 and p < 0.001). This means that a firm’s choice of global CSR pronouncements is determined by both industry and country characteristics. The second stage model shows that CSR_POLICY is significantly and positively associated with FRAMEWORK, controlling for the predicted FRAMEWORK score from the first stage (coefficient = 0.1799 and p < 0.001). This shows that multiple CSR pronouncements support normative isomorphism that promotes harmonisation of CSR reporting after considering addressing endogeneity concern. Similar to Sun et al. (2020), this study performs three tests to check the validity of our instruments in Table 9. First, the underidentification test, or Kleibergen–Paap rk LM statistic, reveals that instrumental variables are well identified (p = 0.0000). Second, the result of the weak identification test or Kleibergen–Paap rk Wald F statistic (Kleibergen and Paap, 2006) shows that the F-statistic is very high in our sample (72.863), which suggests our instruments are relevant and strong. Third, Sargan statistic or the Hansen’s J statistic (Hansen, 1982) is used to check the over-identification concern. The p-value of the Hansen’s J statistic is high (0.4643), suggesting that we cannot reject the null hypothesis that the instruments are exogeneous. This means that other sources of endogeneity, such as omitted variables or simultaneity issues, are unlikely in the model. Overall, our post-estimation tests confirm both the relevance and the exclusion restrictions of our instruments.

To address selection bias because of unobservable differences, this study uses a Heckman two-stage model (Tucker, 2010). The unobservable self-selection bias could violate the relevance of the main findings because the sample of this study consists of only those firms that voluntarily issue CSR reports. Therefore, to address concerns about selection bias, the Heckman approach is used to develop a probit model for CSR practice level (i.e. GRI/framework score). In the first stage of the Heckman model, this study uses GRI standards as a dummy variable that equals to 1 if the sample firm prepares CSR reports with GRI standards and 0 otherwise (GRI). The sample size in the first-stage model is 2079 firm-year observations. The first-stage model should include a variable that satisfies the exclusion restriction. That is, such a variable should be related to the choice of GRI/framework score but unrelated to CSR_POLICY. Then this study uses the Inverse Mills Ratio estimated from the first-stage model as an additional independent variable in the equation. Our first-stage probit model is specified as follows:

where Env is a measure of environmental performance score for firm i in year t, which is obtained from Eikon. Env is used to satisfy the exclusion restriction, and Env is then converted into country-level, average. Specifically, firms with good environmental performances are likely to face a greater demand for non-financial information, and firms would be more willing to report more regarding their performance (Dhaliwal et al., 2011). These firms would be more likely to use global CSR pronouncements to get the most benefit out of them. However, environmental performance is unlikely to be directly related to harmonisation except when reported through or via CSR reporting pronouncements. We include several variables following Fortanier et al. (2011) and Mazboudi et al. (2020).

Table 10 shows the results of the Heckman two-step model. In column 1, Env is positively and significantly (coefficient = 0.0155 and p < 0.05) associated with a higher level of CSR practice, which is consistent with expectations, suggesting that Env is reasonable exogeneous variables to satisfy the exclusion restriction criteria. Other variables are also significantly related to the CSR level, such as SIZE, CSRCOMMITTEE, AWARDS and DIVERSITY, and the model has a Pseudo R² of 36.67%.

In the second stage, the coefficients of FRAMEWORK remain positive and significant (Coefficient = 0.1010 and p < 0.001), consistent with firms adopting multiple global CSR pronouncements having higher harmonisation that reduces country-of-origin effect. Therefore, the results are consistent after controlling for unobservable differences between high CSR and low CSR firms.

This study examines the harmonisation effects of using multiple global CSR pronouncements on firms’ CSR reporting. This study expects that firms using more global CSR pronouncements will result in smaller cross-country differences and fewer country-of-origin effects in CSR reporting. The research, conducted across 16 EU countries with a total sample of 2,443 firm-year observations, collects data on CSR policies and the use of global CSR pronouncements. The results support our hypothesis: use of more global CSR pronouncements is associated with better harmonisation in CSR reporting across firms from different countries. This is consistent with Fortanier et al.'s (2011) and Mazboudi et al.'s (2020) findings that the power of convergence supersedes the power of local institutions in explaining CSR reporting among firms that have adopted global CSR pronouncements. Consistent with Einwiller et al. (2016), use of more global CSR pronouncements reduces selective reporting in CSR reports, which promotes harmonisation of CSR reports. Furthermore, the findings support Zinenko et al. (2015) and García-Sánchez et al. (2019), who suggest the complementary nature of CSR pronouncements. Endogeneity is a major concern that could adversely affect the relevance of the findings. This study addresses this concern through a lagged analysis, firm fixed effects, instrumental variable analysis and Heckman selection bias test. While none of these tests by themselves can remove risk of endogeneity, combined they increase confidence that the results are consistent with the expectations.

For future studies, it would be relevant to further explore this relationship beyond EU countries. Although the EU represents a large proportion of global market capitalisation and has significant impact on international business, the inclusion of other large economies, such as China, India and the USA, may shed additional light on the relationships examined in this study. Another area that may shed further light on the literature is to cover firms that are smaller, as our study used Eikon, which tends to cover firms with relatively large market capitalisation. Furthermore, our study excludes organisational factors that may explain institutional differences such as corporate culture, visibility and leadership. Considering those aspects would benefit future research by providing more contextualisation of the firms. Prior literature, such as Russo-Spena et al. (2018), points out that as long as CSR reporting remains within a voluntary regime, it will never be a tool for accountability, and there is a need to have more reliability that those reports will reflect the actual CSR operations of firms. To address such issues, extending the analysis of coherence in the CSR disclosure strategy to the different channels (such as Web content and advertising campaigns) would fill the gap in the literature as our study’s method is focused on CSR reporting, and it is outside the scope to consider other channels of CSR disclosure strategy.

From a managerial perspective, use of multiple global CSR pronouncements may improve CSR performance and reporting. The resulting increased access to related networks concerning CSR reporting is likely to increase knowledge transfers on CSR management, implementation and reporting. It is also likely to improve the handling of international demands, hence addressing Mazboudi et al.'s (2020) findings, that firms placed in countries with strong institutional environments (e.g. Sweden) focus on addressing local demands rather than international demands. In addition, stakeholders could check to see if global CSR pronouncements are used by CSR reporting companies for an indication of reliability and harmonisation of CSR reporting. This implication is similar to prior literature such as Russo-Spena et al. (2018) and Matten and Moon (2020).

Recent developments concerning the ISSB and EFRAG provide hope for more unified progress towards harmonised international sustainability standards. This study’s findings offer potential support to such entities by providing empirical evidence on harmonisation of CSR reporting. The setting for our study is perhaps more relevant to the ESRS situation (e.g. EU jurisdiction, GRI base and inclusion of SDGs). Consequently, future studies could build on the results of this study to facilitate comparison of ISSB standards and ESRS. Amel-Zadeh and Serafeim (2018) contend that the lack of quantification and the lack of standardisation governing the reporting of environmental, social and governance information are the greatest challenges investors face in integrating environmental, social and governance information into their decision-making process. The ISSB and EFRAG could use the findings from this paper to support their investments in collaborating widely with other entities involved with CSR pronouncements. This study’s evidence on the role of multiple CSR pronouncements in harmonising reporting should also be of wider interest to preparers and users of CSR reports, as well as to academics, and to other regulators and standard setters.

Data availability: Data are available from the specific databases cited in the text.

List of questions concerning CSR policy adoption:

• Does the firm have a policy to improve employee health and safety?

• Does the firm have a policy to support the skill training of its employees?

• Does the firm have a policy to support career development of its employees?

• Does the firm have a flexible working hours policy?

• Does the firm have a diversity and opportunity policy?

• Does the firm have a freedom of association policy?

• Does the firm have a community involvement policy?

• Does the firm have a business ethics policy?

• Does the firm have a policy to protect customer health and safety?

• Does the firm have a quality management systems policy?

Source(s): Eikon Database