Abstract
In recent decades, as worldwide attention to corporate responsibility increased, the global corporate responsibility (GCR) movement did not converge on a singular governance model nor hybridize into myriad country-specific models. The movement, rather, bifurcated into onerous certification frameworks and more lax reporting frameworks. We examine this ‘governance divide’ in the GCR movement by investigating the cross-national diffusion of seven core GCR frameworks. We adopt a glocalization perspective that conceptualizes a vertical nesting of local and global contexts. Our cross-national quantitative analyses suggest that, while linkages to global culture have encouraged business participation in all GCR frameworks, power dependencies related to international trade and domestic factors related to effectiveness of local governance institutions have contributed to divergent diffusion patterns across reporting and certification frameworks. We discuss these findings in relation to several organizational perspectives and note their implications for further research on corporate responsibility.
The post-war era has been characterized by the dramatic proliferation of global organizational models, one prominent case being global corporate responsibility (GCR) models that encourage businesses to attend to their social and environmental impacts (Lim & Tsutsui, 2012). Yet even as attention to corporate responsibility issues has increased worldwide, the GCR movement has not converged on a singular worldwide framework nor hybridized into myriad country-specific models (Meyer, Boli, Thomas, & Ramirez, 1997; Aguilera & Cuervo-Cazurra, 2004). Instead, the GCR movement has come to reflect a global ‘governance divide’ in the form of two distinct organizational models: reporting frameworks with lax requirements and business discretion in the content of voluntary disclosures, and certification frameworks that comprise costly commitments and detailed standards, often accompanied by auditing and enforcement provisions. In this paper, we examine the factors that have led to the differential diffusion of these two models of corporate responsibility across countries.
In so doing, we propose that research on global organizational governance should explicitly address how structuration, ‘the formation and spread of explicit, rationalized, differentiated organizational forms’ (Meyer et al., 1997, p. 156), engenders not only the increasing density of organizational models but also their increasing pluralism (Pope & Meyer, 2016). We draw on the ‘glocalization’ perspective (Drori, Höllerer, & Walgenbach, 2014), especially its conception of the vertical nesting of local and global contexts, to distinguish two dimensions of global diffusion: (i) an overall global level of diffusion, in which core organizational ideas gain widespread prominence across countries (Meyer et al., 1997) and (ii) a nested level of diffusion that is tied to particular country-level contexts in which organizational ideas are adopted (Jackson & Apostolakou, 2010). This paper investigates these parallel processes in the case of seven core GCR frameworks that aim to regulate the social and environmental impacts of businesses. Our cross-national quantitative analyses reveal that, while the overall increase in the diffusion of private governance models reflects country linkages to the world society, divergences in the diffusion of these models reflects enduring differences in countries’ economic and governance infrastructures. Specifically, trade linkages to the global economy, and the perceived ineffectiveness of domestic governance, are associated with organizational participation in more stringent certification frameworks. Conversely, organizations in countries with more effective local governance have higher rates of participation in reporting frameworks that showcase more discretionary corporate responsibility.
As our main contribution to organization studies, we suggest that attention to this governance divide helps us to understand how many of the governance models that constitute and constrain organizations, although created at the international level, nonetheless diffuse in a manner that partly reflects the nesting of organizations within particular local institutional contexts. Organizations themselves are the adopters of these governance models, although the forces that promote adoption oftentimes lie outside organizational boundaries, in local external environments that are differentially linked to world society and which have their own unique institutional strengths and weaknesses. In the case of GCR models, increasingly global attention to corporate responsibility in the context of very disparate country-level institutional environments has produced neither a convergence nor country-specific hybridization, but a bifurcation of governance models that reflects enduring trade and governance differences between core and non-core regions of the world. This interaction of global and local domains in the case of corporate responsibility has profoundly shaped, not only the diffusion of governance models, but also the participation structures by which organizations express and implement their obligations to society and the natural environment.
The Core GCR Frameworks
Although the GCR movement began with intergovernmental efforts during the 1970s (Sagafi-Nejad, 2008), it gained widespread momentum only in the 1990s with the emergence of a litany of voluntary frameworks that were adopted by companies throughout the world (Segerlund, 2010). In this paper, we analyse seven frameworks that are now core components of a highly developed GCR movement: the Carbon Disclosure Project, Fair Labor Association Code of Conduct, Global Compact, Global Reporting Initiative, ISO 14001 Standards, Principles for Responsible Investment and SA 8000 Standards. Well recognized by practitioners and scholars alike as major GCR frameworks, each has substantial business participation internationally and a history of operations exceeding a decade. Table 1 presents brief biographies of these frameworks; Table 2 operational and analytic differences; and Figure 1 participant counts across time.
Biographic Sketches of Seven Core GCR Frameworks.
Comparison of Analysed GCR Frameworks.
Worldwide Counts of Organizational Participation in GCR Frameworks by Year.
Ours is the first effort to study more than a half-dozen GCR frameworks in a single comparative analysis. While GCR research is growing, empirical studies are rarely designed to make divergences across frameworks a central concern. A survey of the scholarship reveals that 63 percent of the existing 2,217 articles on GCR frameworks focus on just the ISO 14001 Standards (see Table 7 in the appendix). The Global Compact and Global Reporting Initiative receive nearly three times as much scholarly attention as the remaining five frameworks, with the Principles for Responsible Investment largely escaping analysis. The absence of comparative work is especially apparent for the Fair Labor Association Code of Conduct, Carbon Disclosure Project, and Principles for Responsible Investment, where the associated literature consists mainly of framework-specific studies. While the existing literature has been insightful, an explicitly comparative approach is needed to reveal divergences in the diffusion and structures of these frameworks and to investigate how these divergences have contributed to the overall structuration of the GCR movement.
Individually, the core frameworks exhibit many of the hallmarks of organizational models studied by globalization scholars (Bromley & Meyer, 2015). The frameworks have a large number of participants, which in each case has increased dramatically from 1998 to 2017 (Figure 1) and now exceeds 1,000 companies (Table 2). Each framework is iteratively constructed by disparate stakeholders working collaboratively to rationalize developmental problems. The ISO 14001 standards, for example, are the creation of the International Organization for Standardization, a confederacy of 164 national groups and a network of 700 international, regional and national standards boards. Likewise, the Global Compact promotes its standards in consultation with thousands of cities, universities, non-profits and labour groups, any of which can become participants by committing to the framework’s core principles. A similarly broad-based construction process characterizes the other five frameworks.
Supporting their status as global governance models, the GCR frameworks are endorsed by prominent international organizations. The United Nations (UN) has been particularly supportive (Ruggie, 2003), itself administering the Global Compact and the Principles for Responsible Investment. The UN also partners with the MBA Oath (which targets MBA students) and the Global Reporting Initiative; oversees the Principles for Responsible Management Education; and lends expertise to other major frameworks, including the ISO 26000 and the European Eco-Management and Audit Scheme. Another influential intergovernmental organization is the Organization for Economic Co-operation and Development (OECD), which helms the Guidelines for Multinational Enterprises, a treaty-like framework signed by nation-states. Lastly, prominent business forums like the World Economic Forum and International Chamber of Commerce have also incorporated corporate responsibility concerns into their various missions (Pigman, 2007).
The GCR frameworks feature extensive cross-national organizational participation, often from the largest transnational corporations, powerful global actors in their own right. A total of 93 percent of the top 250 corporations in the world by revenue participated in the Global Reporting Initiative in 2017 (KPMG, 2017). Likewise, about 80 percent of the Fortune 500 had participated in the Carbon Disclosure Project’s cross-national survey by 2009 (Carbon Disclosure Project, 2010). Many major corporations participate in multiple GCR frameworks. In 2016, for example, Coca-Cola, Cisco, Dow Chemical, Ford, Intel, Tata Motors and other multinational firms were simultaneously participants in the Global Compact, Global Reporting Initiative and the Carbon Disclosure Project.
Finally, these seven GCR frameworks are well recognized in scholarly and practitioner discourse. Nearly all prominent handbooks, codebooks, encyclopedias and directories that attempt a comprehensive survey of the GCR field make mention of nearly all frameworks (see Table 8 in the appendix). These frameworks, as such, are generally perceived to be part of a cohesive GCR movement, despite the differences among them that we detail below.
Two Models of Global Corporate Responsibility
Notwithstanding the overall worldwide rise of numerous GCR frameworks, observers have noted divergences among frameworks in how they address nationally and regionally oriented issues and in where they diffuse (Gilbert, Rasche, & Waddock, 2011; Jamali, 2010; Mueckenberger & Jastram, 2010; Rasche, 2009). Notably, businesses in different countries appear to be drawn to frameworks with differing participation requirements and degrees of stringency. Reporting-based frameworks with low implementation costs, limited monitoring provisions and few enforcement mechanisms tend to diffuse to multinational corporations in developed countries (Smith, 2010; Whitehouse, 2003). At the same time, smaller factories in developing countries tend to be the adopters of frameworks that feature extensive rules, costly implementations and enforcement through intrusive audits (Bartley, 2007b; Locke, Rissing, & Pal, 2013).
Reflecting these different designs and diffusions, GCR frameworks have been subject to two sets of criticisms. Observers have charged that international businesses support reporting frameworks in order to co-opt influential policy groups, gain regulatory and institutional capture, and thereby legitimize lax and non-binding approaches to environmental and social challenges (Smith, 2010; Utting & Zammit, 2008). These criticisms have been vociferous even within the UN (Utting, 2005), with accusations that businesses with poor responsibility records are joining UN initiatives to ‘bluewash’ their images (Bigge, 2004), that is, to mislead stakeholders into believing that their social practices have been sanctioned by international bodies (Berliner & Prakash, 2015). As for certification frameworks, a second set of criticisms targeted the GCR movement’s relative neglect of labour concerns in developing countries. According to these criticisms, factories in developing countries were not themselves entirely responsible for exploitative labour practices since these factories were heavily conditioned by cost-cutting pressures from transnational buyers in developed countries (Bloomfield, 1998). One response was the idea of supply chain monitoring, that transnational corporations that purchase labour-intensive goods from factories in developing countries are also obliged to monitor the manner in which those goods are produced (Locke et al., 2013). Another prominent concern was ‘pollution havens’ or the ‘race to the bottom’, the notion that transnational corporations are incentivized to seek out production partners in countries with the lowest labour and environmental standards (Prakash & Potoski, 2006).
Certification vs. reporting frameworks
Certification frameworks are structured around third-party auditing of corporate responsibility standards at the factory, supplier, or facility level. Auditors for some frameworks may visit the workplace unannounced with a checklist of items related to social responsibility practices and procedures. Guidelines are often extensive, concrete and industry specific (e.g. the Marine Stewardship Council), and participation is relatively costly. By contrast, reporting frameworks are structured around voluntary disclosure and target a broader spectrum of businesses. Companies here may self-declare their greenhouse gas emissions, as in the Carbon Disclosure Project, or self-report their discretionary activities that advance the principles of the Global Compact. Guiding principles of reporting frameworks are usually couched in general and abstract terms and may be so broad as to apply to any company in any industry, or even to non-profits, such as cities and universities, which can also participate in the Global Compact, Carbon Disclosure Project, or Global Reporting Initiative. Participation costs are minimal as they are typically free (Carbon Disclosure Project), recommended (in the first iterations of the Global Compact), or a small percentage of company revenue (Global Reporting Initiative).
These characteristics suggest two distinct groups of GCR frameworks: the Global Compact, Global Reporting Initiative, Carbon Disclosure Project and Principles for Responsible Investment, on the one hand, and the SA 8000, Fair Labor Association Code of Conduct and ISO 14001 Standards, on the other. These distinct groups are evident in existing research. For example, Bartley (2003, 2005, 2007a, 2007b) compares codes (e.g. Fair Labor Association Code of Conduct and the SA 8000) and industries (e.g. forestry and textiles) but addresses only certification schemes. Likewise, other authors compare the AA 1000 with the SA 8000 (Göbbels & Jonker, 2003) or the EMAS with ISO 14001 (Freimann & Walther, 2002), all of which are certification frameworks. For the second group, Lim and Tsutsui (2012) and Shanahan and Khagram (2006) compare the Global Compact and Global Reporting Initiative, both reporting frameworks, while Pope and Lim (2017) add the Carbon Disclosure Project, another reporting framework. While these studies generally discuss their frameworks as representative of the broader GCR movement, these studies in actuality lack the scope to make differences across major groups of frameworks a central feature of analysis. In our view, comparisons across the two major GCR models are needed to achieve a greater understanding of not only the overall expansion of the GCR movement but also its internal pluralism. We concur with Rasche (2009, p. 193) that, despite their differences, GCR frameworks all work toward a ‘common aim’, and thus ‘it seems appropriate to treat these standards as a joint category, at least for the sake of developing a framework for their analysis’.
Theory and Hypotheses
What explains these parallel movements – the overall global increase in participation in GCR frameworks and also the differential diffusion of the reporting and certification models? We suggest that glocalization processes, which point to the ‘mutually constitutive character of the global and the local’ and the ‘dualities of similarity and variation’ (Drori et al., 2014, p. 85; Robertson, 1995; Roudometof, 2016), are key for understanding this GCR governance divide. We emphasize particularly what Drori et al. (2014) refer to as the vertical axis of glocalization, which views diffusion as a process that reflects the nesting of local contexts within the broader global domain, giving rise to ‘manifold ideas, structures, and practices in the form of regulations, standards, or norms’ (Drori et al., 2014, p. 93). Thus, we distinguish between the overall structuration of the GCR movement and the multiple nestings within the broader global canopy. In the following hypotheses, we consider how different glocalization factors, including the role of international organizations, local business associations, economic linkages and countries’ governance institutions shape the diffusion of certification and reporting frameworks.
Overall structuration of the GCR movement
Since the end of World War II, the creation and spread of organizational models has been a ubiquitous feature of globalization (Bromley & Meyer, 2015). In the absence of a global regulatory hegemony, transnational epistemic communities such as universities, professional bodies, scientific associations and social movements have become major sources of global influence (Meyer et al., 1997), prescribing models of organizational practice for a wide range of issues, including the natural environment (Frank, Hironaka, & Schofer, 2000) and human rights (Hafner-Burton & Tsutsui, 2005). This worldwide structuration is also evident in the GCR movement, which has promulgated voluntary governance frameworks for businesses to address their social and environmental impacts (Lim & Tsutsui, 2012; Pope & Lim, 2017). Although many existing approaches to GCR characterize it as an outgrowth of domestic contexts (Jackson & Apostolakou, 2010; Matten & Moon, 2008) or of narrowly conceived strategic business interests (McWilliams & Siegel, 2000), the expansion of GCR is also attributable to wider global processes (Pope & Meyer, 2015).
In particular, country-level linkages to world society via international nongovernmental organizations (INGOs) (Boli & Thomas, 1997) may promote the uptake of GCR frameworks. World society research theorizes INGOs as conduits for the spread of global norms to domestic settings, either through direct pressure on companies (King, 2008) or by fostering companies’ knowledge about GCR practices (Kell & Levin, 2003). Indeed, INGO memberships have been correlated with participation in the reporting GCR frameworks (Lim & Tsutsui, 2012), although an extension to the certification frameworks has remained unexplored.
In addition, some countries may have greater numbers of ‘receptor sites’ for GCR frameworks – ‘social structures … with the capacity to receive, decode, and transmit signals from world society to national actors’ (Frank et al., 2000, p. 96). Due to these capacities, receptor sites have been theorized as increasing the diffusion of global models into local settings (Cole, 2017; Frank, Longhofer, & Schofer, 2007; Frank, Robinson, & Olesen, 2011; Meyer, 2010; Schofer, Hironaka, Frank, & Longhofer, 2012; Suarez & Bromley, 2012). For GCR frameworks, we propose that particularly salient examples of receptor sites are domestic business associations oriented around corporate responsibility. Such business associations tend to have upward connections to the global domain in the form of memberships from prominent multinational corporations. Additionally, many of these business associations are themselves federated into regional business associations such as ASEAN CSR, CSR Europe, and Arabia CSR Forum (Pope & Lim, 2017). Still others are explicit creations of business associations that are global, such as the approximately 70 ‘local networks’ of the Global Compact.
As meso-structures operating at the interface of global and local levels (Kinderman, 2015), business associations such as the local networks of the Global Compact explicitly seek to ‘help companies understand what responsible business means within different national, cultural, and language contexts’. 1 Similarly, the more than 60 nationally focused business councils of the World Business Council for Sustainable Development seek ‘to create the global-local linkages necessary to promote business solutions for a sustainable world’. 2 Domestic business associations, we underscore, also generally operate as ‘knowledge practitioners’ (Drori et al., 2014, p. 92). In the United States, for example, associations like Business for Social Responsibility, 3 the Coalition for Environmentally Responsible Economies 4 and the Boston College Center for Corporate Citizenship 5 perform such activities as organizing conferences and knowledge-sharing platforms, providing consulting services, and publishing research reports and briefs. Domestic business associations, then, are not only structurally positioned to be conduits of diffusion (Suarez & Bromley, 2016, p. 145) but are also structured operationally to assist companies in implementation through the local translation of global norms (Meyer, 2010, pp. 12–13). 6
H1a: Countries with more linkages to world society are more likely to have higher levels of business participation across GCR frameworks.
H1b: Countries with more corporate responsibility business associations are more likely to have higher levels of business participation across GCR frameworks.
Bifurcation in GCR diffusion
The increasing global density of GCR frameworks, as well as their increasing numbers of business participants, suggests the emergence of a worldwide movement around corporate responsibility, and perhaps a degree of policy convergence across organizations in disparate countries (Aguilera & Cuervo-Cazurra, 2004). On the other hand, cross-national studies have noted that complexity, rather than convergence, has increased in recent decades (Alter & Meunier, 2009) in the form of ‘partially overlapping and nonhierarchical institutions governing a particular issue-area’ (Raustiala & Victor, 2004, p. 279). This complexity has been explored in global regimes that exist for trade (Davis, 2009), human rights (Hafner-Burton, 2009) and climate change (Keohane & Victor, 2011). As illustrated above, the GCR movement, as well, now encompasses a diversity of frameworks, which have disparate membership provisions, but also considerable overlap in substantive focus.
Consistent with the glocalization perspective, even as world society processes promote the overall structuration of the GCR movement, we expect more nested and enduring country-level differences in trade and domestic governance to give rise to divergences in GCR framework diffusion. According to studies of cross-national policy diffusion, the differential manifestation of GCR frameworks in different national contexts is not incidental (Greenhill, Mosley, & Prakash, 2009; Mosley & Uno, 2007) but is conditioned by the existing effectiveness of domestic governance institutions (Jackson & Apostolakou, 2010), as well as by expectations of business responsibility from international trading partners (Cao & Prakash, 2010). At the domestic level, businesses in countries with historically weak regulatory structures may be more likely to adopt onerous external certification schemes in order to facilitate economic exchange with businesses in countries with higher sourcing standards.
More specifically, a greater density of cross-national economic transactions with developed countries may lead to higher levels of business participation in certification frameworks due to direct or generalized pressures for supply chain monitoring (Potoski & Prakash, 2005; Prakash & Potoski, 2006). Supply chain monitoring involves the formal or informal oversight by companies of their non-local suppliers to ensure that they adhere to specific responsibility standards. For efficiency and external legitimacy, supply chain monitoring is commonly enacted through membership in third-party frameworks, whose administrators are delegated the authority to build consensus on responsibility standards that are mutually acceptable to purchasers, suppliers and nongovernmental groups and to enforce the standards through schemes of certification and auditing (Bartley, 2005; Locke, Amengual, & Mangla, 2009).
As the notion of supply chain monitoring has become institutionalized, companies in developed core countries have come under pressure from ethical consumer segments to create or join certification frameworks and thereby to set in motion processes that lead, for example, to minimum age or minimum compensation requirements at foreign factories (Locke et al., 2013). The Fair Labor Association and the SA 8000 are specific frameworks in our sample that inhere the logic of supply chain monitoring. While these frameworks tend to be championed by organizations in wealthy countries (e.g. prestigious US universities whose students mobilized in the 1990s to express concerns about the manufacturing conditions of university apparel and licensed products), the ground-level enforcement of such frameworks tends to occur in non-core countries that have many facilities involved, for example, in labour-intensive goods for export (Locke et al., 2009; Riisgaard, 2009; Rodriguez-Garavito, 2005; Yu, 2008). In recognition of this direction of cross-national dependency, we expect factory-level participation in the more stringent certification frameworks to be more strongly associated with organizations in countries where there are more economic transactions with, specifically, developed countries.
H2: Countries with more economic linkages to developed countries are more likely to have higher levels of business participation in certification frameworks.
Lastly, in addition to cross-national economic pressures, the quality of domestic governance institutions may shape how global corporate responsibility concerns are perceived and mobilized around (Meyer & Höllerer, 2010). More broadly, business social responsibility may be endogenous to countries with extensive existing legal and social support for human rights, environmentalism, labour and transparency (Vogel, 2010). Addressed to this claim, Jackson and Apostolakou (2010) propose that corporate responsibility may either mirror a country’s strong local governance institutions or substitute for the lack thereof. In the case of GCR frameworks, we consider both sides of the mirror/substitute debate (Brown & Knudsen, 2013; Burgoon & Fransen, 2017), given that existing research has found support for the expansion of corporate responsibility in both poorly (Kaplan, 2015; Kinderman, 2012; Marens, 2012) and effectively regulated national environments (Favotto, Kollman, & Bernhagen, 2016; Midttun, Gautesen, & Gjølberg, 2006). Consistent with institutionalist approaches to organizational research (Bromley & Meyer, 2015), we envision mirroring and substitution, not merely as functionalist processes, but as deeper cultural phenomena whereby organizations seek to make themselves isomorphic with prevailing domestic or global norms.
On one hand, consistent with the substitution thesis, we expect more business engagement with onerous forms of corporate responsibility, such as the certification frameworks, in countries with less effective governance institutions. Businesses operating in environments with weak social and environmental protections may realize competitive advantages over their local peers by adopting forms of GCR that are codified, stringent and certifiable. These advantages may stem from greater access to foreign trading partners through the demonstration of substantive attention to corporate responsibility (Cao & Prakash, 2010). In less effective governance contexts, lax reporting frameworks might not be perceived as providing the necessary degree of external assurance of labour and environmental management. Conversely, businesses in countries with more effective governance institutions may have little need for third-party monitoring of production processes beyond existing regulatory dictates. For the certification frameworks, their potential redundancies with the already strong regulations of countries like Sweden, Germany and Canada may limit diffusion, especially given that such countries have long moved beyond issues related to child and forced labour to encompass more progressive matters such as maternity leave and sexual identity discrimination.
On the other hand, the mirror hypothesis states that businesses may be more likely to adopt some forms of corporate responsibility where there is congruence with existing and generally robust governance institutions. For the reporting frameworks, we underscore that they operate primarily by a logic of transparency, which is the life blood of good public governance and which is a deeply cultural asset that is very difficult to increase merely through legislative fiat. 7 The reporting frameworks require companies to reveal their internal workings through reports that are publicly stored in online databases, which ordinary citizens can download, read and mobilize around. This logic of transparency that underpins the reporting frameworks informs our expectation that these frameworks will diffuse much more readily to countries with effective local governance.
For the mirror hypothesis, we also emphasize that public and private governance are not dichotomous as much as endogenous to one another. It is generally true, in other words, that ‘the stronger social regulation of the economy facilitates … stronger and more extensive private regulation’ (Kinderman & Lutter, 2018, p. 2). Indeed, the effectiveness of both public and private governance is likely to rest on similar institutional foundations, not only transparency, but also a free press, an educated and informed populace, and societally shared capacities for organizing and monitoring. A classic study illustrating the endogeneity of private and public governance is Edelman, Uggen and Erlanger (1999), which chronicled the development of internal ‘grievance procedures’ by American companies in the 1980s as an in-house means of compliance with ambiguous sexual harassment laws – procedures that eventually achieved legal standing before the courts as suitable and sufficient.
The endogeneity of public and private governance is quite apparent in the case at hand in the reporting frameworks. In the European Union, for example, where public regulation is comparatively strong, the European Commission has proactively developed an explicit strategy for improving the regulation of local companies by promoting and supporting the corporate responsibility movement (Aßländer, Gössling, & Seele, 2016). Over time, the European Union has recognized some of the movement’s private voluntary frameworks as a means of achieving public regulatory compliance. For example, as of 2018, companies in the European Union with more than 500 employees are legally mandated to issue corporate responsibility reports and are encouraged to do so by using frameworks like the Global Reporting Initiative. The Carbon Disclosure Project is another example. Although a private framework, participation has been spread through traditional, highly legalized tools of corporate governance, with hundreds of shareholder resolutions being lodged in recent decades by investors seeking to compel participation in the framework. 8 These efforts have attained a degree of success, as companies in the United Kingdom, for example, are now legally required to report on their carbon emissions. 9 Overall, our argument is that the reporting frameworks may be at the vanguard of new and more progressive forms of regulation in countries that already have strong public governance.
H3a: Countries with less effective governance institutions are more likely to have higher levels of business participation in certification frameworks.
H3b: Countries with more effective governance institutions are more likely to have higher levels of business participation in reporting frameworks.
Data and Methodology
The sample for our analyses comprises the 193 countries that are UN member-states, plus Taiwan. The dependent variables are country-level annual business participation counts in the seven GCR frameworks, tracked for each framework for all available years of data from 1998 to 2017. At the organizational level, participants are counted as framework signatories (e.g. the Global Compact), producers of the framework’s annual report (e.g. the Carbon Disclosure Project), or holders of an active certification (i.e. the certification frameworks, with the exception of the FLA Code, which publicly releases only the number of annual country-level random audits).
For the certification frameworks, we used a conventional indicator of country linkages to world society (Boli & Thomas, 1997; Frank et al., 2000; Schofer & Hironaka, 2005), the count of INGOs in which countries have annual representatives. We collected this indicator from various years of the Union of International Associations’ (UIA) Yearbook as organizations that are categorized as ‘International Organizations’ (Cluster I organizations in groups A, B, C, D and F). For reporting frameworks, we were able to make the INGO measure more GCR-specific by counting the number of non-business organizations in a country that have supported the GCR movement through membership or participation in a focal GCR framework. As noted previously, not-for-profit organizations such as labour groups and trade associations can, for example, join the Global Compact or produce social responsibility reports using GRI guidelines. Similarly, cities can report emissions through the CDP Cities Program, and organizations that provide technical support and investment advice to asset managers can become signatories to the Principles for Responsible Investment. Thus, for each reporting framework, we counted non-business participants to indicate the number of civil society organizations in the domestic context that are specifically oriented to world society discourses surrounding the respective framework. To indicate sites for the glocalization of social responsibility, we counted the number of national business associations that are oriented around corporate responsibility. Among these associations are the ‘local networks’ of the Global Compact. They also include the local councils of the World Business Council for Sustainable Development in addition to other standalone responsibility-themed business associations (e.g. CSR Sweden, Canadian Business for Social Responsibility), which we retrieved primarily from Grayson & Nelson (2013), but also from other GCR field surveys, such as those listed in Table 8 in the appendix.
For country linkages to the world economy, we employed several measures that reflect the industrial scopes of particular frameworks. For the FLA and SA 8000, both of which target factory labour in small and medium production facilities, we gathered the dollar value of labour-intensive exports specifically to developed countries as reported in the online database of the UN Conference on Trade and Development (UNCTAD). For the ISO 14001, where participation now encompasses many industrial sectors, we gathered the total value of exports and services specifically to developed countries across all industries as reported by the International Monetary Fund (IMF). Turning to the reporting frameworks, for the Principles for Responsible Investment, which targets banks, asset managers and financial service providers, we gathered the total value of exports worldwide of financial services (UNCTAD). For the remaining initiatives (the Carbon Disclosure Project, Global Compact and Global Reporting Initiative), where membership, like the ISO 14001, tends to cut across sectors, we also used the total value of exports and services worldwide across all industries as reported by the IMF.
To measure the quality of local governance, we employ the government effectiveness score from the World Bank’s database of world governance indicators. This score reflects a government’s ‘quality of policy formulation and implementation, and the credibility of the government’s commitment to such policies’ (Kaufmann, Kraay, & Mastruzzi, 2011, p. 223), as assessed by perceptions data from households, firms and experts. The score is reported in units of a standard normal distribution and is based on over 30 data sources, including Gallup, Reporters Without Borders, the US State Department and the World Bank.
We include numerous controls for the local institutional context. At the level of individuals, we incorporate the Human Development Index of the UN Development Programme, reflecting citizens’ health and longevity, years of schooling, and standard of living as indicated by per capita gross national product. To control for the nature of the local governmental system, we employ the Freedom House political liberties index to capture the extent to which a government resembles a liberal democracy through such attributes as free, open and competitive elections as well as minority representation in government. To control for the nature of the local economic system, we include the Heritage Foundation economic freedom score to reflect the degree to which a local economy resembles a free market, by scoring highly on such sub-dimensions as free trade, limited government spending and property rights. We control for the size of the national economy as the logged USD value of gross domestic product from the World Governance Indicators. To control for the endorsement of world society principles by nation-states, we included the total number of multilateral UN treaties ratified by the nation-state. As for other prominent linguistic, religious and legal dimensions of cross-country institutional variation, we control for English as a de jure or de facto official language or a widely spoken second language (Liang, Marquis, Renneboog, & Sun, 2014); the percentage of the population identifying as Protestant using the most recent figures from the CIA World Factbook (see Brammer, Williams, & Zinkin, 2006 for a rounded discussion of religion and corporate responsibility); and a binary variable for a civil law legal tradition (our rationale here reflects Liang & Renneboog, 2017).
We also add controls for businesses that have been historically targeted for participation in each framework. For the FLA Code of Conduct and the SA 8000, we count the number of businesses within each country in the entire North America and Global files of the Compustat database that are classified as apparel manufacturers (SIC code 23). For the ISO 14001 standards, Global Compact and Global Reporting Initiative, where participation is increasingly cross-sectoral, we include a count of all domestic listed companies from the World Bank. For the CDP, which in its first few years was administered only to the largest companies in the world, we count the number of country-level Compustat companies within the top 1,000 worldwide in annual revenues. For the Principles for Responsible Investment, we control for all Compustat financial companies (SIC codes 61–64). Since the initiatives exhibit a clear upward trend in membership in recent decades, we include year dummy variables. Finally, acknowledging that world regions have unique endowments of natural resources, geopolitics and institutional legacies, we add controls for nine groupings: Asia; Australia and South Pacific; Central America and the Caribbean; Europe (North and South); Europe (Eastern); North Africa and the Middle East; North America; South America; and Sub-Sahara Africa.
All independent variables are lagged by one year as a guard against simultaneity. We apply a natural logarithmic transformation to independent variables that exhibit a skewed distribution. Models are parameterized by the negative binomial distribution, since dependent variables are overdispersed (i.e. variances exceed means) (Hilbe, 2007). Diagnostics for multicollinearity reveal that only GDP consistently has a variance inflation factor larger than 10. Dropping GDP substantially reduced the collinearity problem but, since it does not meaningfully change the pattern of results, we retained GDP in all models. We use the cluster-robust sandwich estimator of standard errors to relax the assumption of no intra-country autocorrelation in the idiosyncratic error term and to guard against arbitrary heteroskedasticity. Diagnostic tests for outliers do not reveal concerns, in part because variables with skewed distributions have been logarithmically transformed. Results are broadly resistant to the deletion of large countries (e.g. China, USA, India), to the deletion of countries with the most participants in a given framework (e.g. Taiwan, a smaller country, produced the most GRI reports in 2015) and to the Winsorization of dependent variables at various cut points.
Results
Tables 3, 4 and 5 report descriptions, sources, univariate statistics and correlations for the variables used in the analysis. Table 6 presents the findings from our negative binomial regression analyses. 10 Tables 9 and 10 in the appendix report membership data across all frameworks by world region and by the largest 50 countries in the world by GDP. Table 11 displays results with standardized coefficients for predictor variables to assist in the comparison of coefficient magnitudes.
Variable Descriptions.
Univariate Statistics of Key Variables.
Note: Variables are unlogged to ease interpretation of magnitudes.
Pair-wise Correlations.
p < .01. Variables are logged where appropriate to match their treatment in the models. To fit this table on a single spread, we do not report correlations for the non-business participants in the various reporting frameworks.
Negative Binomial Panel Regressions of Annual Participation Counts in the Major GCR Frameworks.
The results show that connections to world society have positive, significant effects on participation counts in all frameworks, whether these linkages are indicated by INGO memberships in a given nation or by domestic organizations that have explicitly supported a focal GCR framework. This pattern provides robust support for H1a. For H1b, the association between business associations and GCR framework participation is uniformly positive in all models and statistically significant in 5 of 7, the exceptions being the models predicting participation in the Fair Labor Code of Conduct and the SA 8000. This provides somewhat consistent evidence for the argument of H1b that organizations in countries with dedicated associational venues for the glocalization of GCR have higher counts of participation in multiple types of GCR frameworks. Considered alongside the findings for H1a, this pattern suggests that participation in GCR frameworks may be increased by many types of linkages to world society, whether international nongovernmental organizations, national organizations that express explicit support for international frameworks, or national associational venues that are likely to be receptor sites for international norms. These results also provide some validation to the world society perspective’s argument that global cultural factors have significant influences that are independent of ‘functionalist’ factors. In Table 11 in the appendix, which presents results with standardized regression coefficients, variables that indicate cultural factors, depending on the model and the comparison functionalist variables (governmental effectiveness or values of exports), oftentimes have higher magnitudes.
Turning now to country linkages to the world economy, we find that the value of exports has positive and significant effects (p < .01) on all three certification frameworks (the FLA Code of Conduct, ISO 14001 and SA 8000). This supports H2, suggesting an association between international trade with developed countries and frameworks that have stringent, explicit, and relatively costly membership requirements. We note that, in the models predicting participation counts in the SA 8000 and Fair Labor Association Code of Conduct, the standardized coefficients for the trade variables are the largest of all variables (Table 11 in the appendix).
We find that the perceived quality of domestic governance has different effects on counts of framework participation depending on whether the framework operates through a certification or reporting logic. Weak domestic governance associates with increased participation in the certification frameworks (except ISO 14001), whereas the opposite is generally true for the reporting frameworks. These findings generally corroborate H3a and H3b. This pattern is compatible with the argument that certification frameworks may be used by organizations to overcome perceived weaknesses in domestic governance and thus may serve as institutional substitutes. Meanwhile, in contexts where local governance has high perceived effectiveness, reporting frameworks may provide businesses a platform to signal their beyond-compliance and discretionary approaches to social responsibility without incurring the financial or administrative costs of certification.
As just mentioned, results for the model predicting ISO 14001 participation contradict H3a, perhaps indicating that participation in this framework does not increase with the effectiveness of domestic governance. To delve deeper, we ran additional models that included a squared term for governmental effectiveness to investigate whether we would observe nonlinear effects of governmental effectiveness on participation in GCR frameworks. Results for these models are displayed in Table 12 in the appendix. The models show that for ISO 14001 the unsquared term is positive and significant and the squared term is negative and significant. Inserting values of the governmental effectiveness score at various points in its distribution, while assuming other variables are held constant at their means, reveals that predicted participation counts rise in ISO 14001 as governmental effectiveness increases, but begin to fall for companies with a governmental effectiveness score of more than two standard deviations higher than the mean (in countries such as Denmark, Finland and Singapore). Similarly, predicted membership counts in the Carbon Disclosure Project and Global Reporting Initiative begin to slow or decline as governmental effectiveness reaches its highest levels. These patterns add some nuance to our results, suggesting that companies in very high governmental effectiveness contexts may be insulated from having to participate in some GCR frameworks. We note, however, due to the non-significance of the unsquared term for predicting participation in the Fair Labor Code and the SA 8000, that the relationship of these initiatives with governance effectiveness is probably best modeled as linearly negative.
Additional analyses and robustness checks
We performed the following additional analyses. First, we ensured that the main results did not change for alternative controls of religion, language and legal system. Second, we excluded recently formed INGOs from analysis, since their newness might limit their ability to influence GCR participation. Third, we used seemingly unrelated regression techniques to rule out concerns related to the correlation of error terms. Fourth, we specified analogous models with organizational- rather than national-level units of analysis. We did not specify these as main models because facility-level participants are not released for some frameworks (ISO 14001), because the requisite financial data is not available in financial databases for participants in some frameworks (the SA 8000 or Fair Labor Code), or due to concerns about statistical power (only 55 of the participants in the Principles for Responsible Investment could be located in the Compustat database). In these company-level models, in place of world economy linkages, world society linkages and governmental effectiveness, respectively, we used a company’s total value of sales generated abroad (Thomson Reuters database); company-level memberships in major global business associations oriented around social responsibility, namely, the World Business Council for Sustainable Development, International Business Leaders Forum and the World Economic Forum (following Pope & Lim, 2017); and a company’s corporate governance score from the Asset4 database. We included controls for organizational size and slack resources, and dummies for year, country and industry. These additional analyses (Table 13 in the appendix) are generally congruent with our main findings, with world society linkages and corporate regulatory strength having a positive association with participation in the reporting frameworks, whereas international trade has no observed significant relationship.
Discussion and Conclusion
The imagery that follows from the above findings is of a thoroughly penetrative global cultural domain that has spread the corporate responsibility movement to the far corners of the world, but that has nonetheless diffused distinct governance models to address situated problems, many of which appear to have stemmed from the competitive pressures of international trade.
How do we make sense of these parallel forces in the worldwide GCR movement? In light of our findings, we point out that it is analytically difficult to conceive of institutional pluralism without also acknowledging the overall worldwide structuration that affords such pluralism in the first place. Pluralism in the development of distinct organizational models of corporate responsibility derives from the initial emergence of voluntary and lenient standards for regulating the social and environmental practices of corporations. Over time, even as the GCR movement gained adherents and influence, consensus formed around the inadequacy of the initially weak sanctions against irresponsible business actors. The movement subsequently addressed these emerging criticisms and the exigencies of specific local contexts by developing differentiated organizational models with unique expectations and standards. As this occurred, the GCR movement did not converge on a singular worldwide framework, neither did it generate a profusion of models catered to specific recipient countries (Aguilera & Cuervo-Cazurra, 2004). Rather, the movement came to exhibit a governance divide by promulgating two major types of GCR models whose diffusions reflected enduring differences between core and non-core regions of the world (Chase-Dunn, 1999).
We fully anticipate that the governance divide in the GCR movement will continue to drive much dynamism and change in the movement’s development and further its worldwide expansion. Indeed, the contradiction in contemporary global corporate responsibility in which the most powerful companies participate in the most lenient initiatives has already led to numerous reforms. The Global Compact, for example, has transitioned from an extremely lenient initiative to one with increasingly onerous membership requirements. Similar rounds of elaboration and reform have occurred for the Carbon Disclosure Project and the Global Reporting Initiative, both of which are now past their third iteration and have incorporated new integrity measures to encourage substantive participation. Nevertheless, we also acknowledge that regime complexity may favour more powerful actors with the leverage to shape global institutions in their interests (Alter & Meunier, 2009).
We see our study as having several main contributions. First, our findings emphasize that external contexts may be significant drivers of the governance models that constitute and constrain organizations. The GCR movement is a particularly significant case – its prescriptive models of organizational practice, whether lax or stringent, appear to be determined by actors and forces far beyond just businesses themselves. GCR frameworks have now diffused rapidly and widely, signalling a new phase in the development of social and environmental accountability that cannot be explained solely by national or industry-level demands (cf. Kinderman, 2013). In the global context, conspicuously lacking a regulatory hegemony, the GCR movement has operated in a neoliberal and decentralized fashion, developing pluralistically while nonetheless professing to advance the same slate of universal and abstract governance principles (Meyer, 2010). Our paper suggests that the structuration of the worldwide GCR movement is a dynamic, compound process (Boyle, Kim, & Longhofer, 2015; Hironaka, 2014) with both an expansionary trajectory (reliant on common and abstract principles) and pressures towards pluralism (frameworks tailored to core and non-core contexts).
Second, our findings indicate that globalization studies need to be attuned to the divergent rather than convergent diffusion of some organizational models. The glocalization perspective that we employed alerted us to nested levels of diffusion that contributed to a governance divide in GCR models. Similarly, in other substantive areas, glocalization studies have found that even the same global model can permit striking heterogeneity in practices across countries (Drori et al., 2014). Our findings highlight how factors such as domestic governance and country linkages to the world economy can generate differential diffusion within the context of overall institutional expansion. Future studies could further explore this dynamic in world society, paying attention to how enduring political and economic differences between countries generate heterogeneity within a widening canopy of global institutional structures. Along these lines, researchers could continue to explore how national organizations (e.g. business associations) assist in the production of global models, increase their diffusion to national settings (Kinderman, 2015) and translate the principles of those models to facilitate substantive implementation within particular domestic contexts (Drori et al., 2014; Meyer & Höllerer, 2010).
Like Stetter (2007) and Hironaka (2014), our study contributes to the discussion of how power and politics shape global outcomes. Globalization perspectives like the world society perspective have been criticized for de-emphasizing power relations (Buhari, 2010), including the vast differences in countries’ material and relational resources (Beckfield, 2003). Our findings suggest that these power imbalances have resulted in an outcome where lenient GCR frameworks apply mostly to companies in countries with the need to legitimate their outflowing capital interests while the onerous GCR frameworks appear mostly in countries with a high amount of labour-intensive exports. These power imbalances have apparently also impacted the membership provisions of the GCR frameworks (Smith, 2010), given that the concerns surrounding corporations in highly developed countries, including capital flight, consumer privacy and data misuse, offshore tax evasion, and bribery and corruption, could also be addressed, in theory, by frameworks that require equally extensive and expensive auditing by third parties. We suggest that glocalization approaches, centred as they are on the interactions between the global and the local, will continue to be useful in the analysis of power imbalances as these approaches direct analysts to consider the embeddedness of core and non-core countries within a larger world system.
Third, our study adds to the literature on organizational accountability and responsibility in several significant ways. Our study has delineated different dimensions of the structuration of the GCR movement as well as different channels by which this structuration has occurred, encouraging future researchers to consider both the increasing influence of GCR models and their simultaneous internal differentiation into frameworks focused on more specific geopolitical and economic issues. Our study builds on previous cross-national studies on GCR (Lim & Tsutsui, 2012; Prakash & Potoski, 2006) that, while very insightful, have examined only a narrow range of GCR frameworks. Our findings underscore the importance of integrating a wider scope of core GCR models into a single comparative framework (Behnam & MacLean, 2011; Rasche, 2009), including models that have largely escaped scholarly attention, such as the Principles for Responsible Investment. In taking this larger vantage, we were able to observe glocalization processes that appear to have engendered a governance divide in GCR frameworks, a finding that would have been obscured if fewer frameworks were captured within the same analytical lens.
Furthermore, while many previous studies are extant that have compared countries by their corporate responsibility practices, our study extends this literature in a new direction by relating country-level differences to the local adoption of different types of corporate responsibility frameworks. While Jackson and Apostolakou (2010) find that domestic institutions governing the relations between the state and market have a conditioning effect on the corporate responsibility practices that exist within a country, our analyses reveal that these practices are also influenced by the nesting of local actors within the larger global context. That is, global discourses often compel companies in labour-intensive economies to adopt GCR frameworks with requirements that are drawn out in explicit detail. Meanwhile, companies with strong domestic governance institutions are often conditioned to participate at higher rates in frameworks with abstract principles, which give companies greater latitude in interpreting GCR standards and in devising discretionary responses. These differences suggest that whether corporate responsibility is a mirror or substitute for domestic governance is also tied into questions about the type of GCR framework that is adopted (Kinderman & Lutter, 2018). One implication is that the vociferous criticisms of corporate responsibility as mere window dressing may be less applicable to developing countries to the extent that companies there are more prone to adopt GCR frameworks with explicit guidelines for accountability that are enforceable by third parties. On the other hand, window dressing may still be a concern in developed countries, where businesses have continued to advocate for lenient and voluntary frameworks that do not address their own unique problems such as outsourcing, capital flight and offshore tax evasion. Nevertheless, the expansion of GCR frameworks overall indicates a wider international movement that has increasingly legitimated the notion of private business regulation, while experiencing some degree of success in ratcheting up GCR standards (Bromley & Meyer, 2015).
Finally, our findings have implications for field-level analyses of organizational institutions. As accountability and responsibility doctrines continue to diffuse internationally and to catalyse the development of issue-centric fields in very disparate sectors, it is likely that the governance divide will come into sharper relief as the involved organizations debate the utility of specific responsibility frameworks and the contradictions among them (Bebbington, Kirk, & Larrinaga, 2012). Further attention to differences in responsibility frameworks (stringent versus lax) at the field level are likely to also reveal competing logics of organizational responsibility (Greenwood, Raynard, Kodeih, Micelotta, & Lounsbury, 2011) that may exacerbate means–ends decoupling (Bromley & Powell, 2012) as organizations develop more complex structural interfaces and divert greater resources to accountability and responsibility concerns that are increasingly tangential to core operational goals.
The governance divide in the GCR movement suggests an uneven manifestation of corporate responsibility across institutional environments, not only allowing businesses in countries with strong governance mechanisms to reap reputational benefits from outward displays of good corporate governance (Potoski & Prakash, 2005) but also imposing higher expectations and costs on organizations in labour-exporting countries. This divergence comes into relief when multiple frameworks are examined within the context of a larger GCR movement that is shaped by the broader global forces that this paper has sought to highlight.
Footnotes
Appendix
Company-Level Panel Regressions Predicting Participation in the Carbon Disclosure Project and Global Reporting Initiative or Joining the Global Compact.
| Carbon Disclosure Project |
Global Compact |
Global Reporting Initiative |
|
|---|---|---|---|
| Logistic regression | Event-history regression | Logistic regression | |
| Linkages to world society | |||
| Memberships in Global CSR Associations | 3.09*** | −0.08 | 0.72* |
| (1.10) | (0.35) | (0.35) | |
| Linkages to world economy | |||
| Foreign sales (log, USD; Thomson Reuters) | 0.06 | 0.02 | 0.00 |
| (0.04) | (0.03) | (0.02) | |
| Regulatory context | |||
| Asset4 Corp. Gov. score (Thomson Reuters) | 0.05*** | 0.013* | 0.03*** |
| (0.01) | (.006) | (0.01) | |
| Controls Variables (Compustat Global) | Total Revenues; No. of Employees; Total Assets; Current Assets; Current Liabilities; Long–term Debt; Shareholders Equity; and dummies for Year, Headquarter Country, and 2–Digit SIC Code | ||
| Constant | 6.33*** | −8.42*** | −39.01*** |
| (2.12) | (1.45) | (8.02) | |
| Log likelihood | −935.52 | −371.63 | −1442.43 |
| Company-years | 3,113 | 5,197 | 3,581 |
Standard errors in parentheses. †p < .10; *p < .05; **p < .01; ***p < .001 (two-tailed tests).
The starting base sample is the entire Global File of the Compustat database.
The hazard rate in the event-history regression is modelled with year dummies.
Global CSR Associations is count variable ranging from 0 to 3 of total memberships in the World Business Council of Sustainability; the International Business Leaders Forum and the World Economic Forum (the major international CSR associations identified by Pope & Lim, 2017).
The Asset4 Corporate Governance Score is a percentile index ranging from about 1 to 99 that reflects a corporation’s position relative to peers in terms of assessments on various corporate governance issues that are calculated from nearly 70 key performance indicators.
Logistical regressions often have convergence problems when there are many dummy variables, a small ratio of positive to negative outcomes, or sticky dependent variables from year to year. To enable convergence, some control variables were dropped from analysis in some of the models. We note that the general pattern of results above is robust to the choice of variables to drop.
The sample size is much smaller than the company-years in the Compustat database primarily due to merging this database with the Asset4 database (which tracks only the largest 4,000 companies worldwide) and with the ‘foreign sales’ variable (which although a clean measure of our intended construct is not widely reported by companies globally).
We do not include a variable for advertising intensity due to a large share of missing values and owing to a literature review finding that previous studies, contrary to accepted wisdom, have not found any consistent relationship in a multivariate context that runs from advertising intensity to social responsibility (Pope, 2016).
Funding
This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.