Global and local orientation in organisational actorhood: A comparative study of large corporations from Germany, the United Kingdom, and the United States^* Open Access

According to the world society perspective in institutional theory, the supralocal sphere of world society constructs organisations as social actors that are expected to adopt globally diffusing formal templates and ideas of organising (Bromley & Meyer, 2015; Drori, 2008; Drori, Meyer, & Hwang, 2009; Meyer & Bromley, 2013). Simultaneously, however, organisations are expected to comply with the distinct institutional traditions of their local fields (Tempel & Walgenbach, 2007). Such a multiplicity of global and local institutional demands result in what is described as ‘glocalisation’ or ‘glocality’ (Robertson, 1995; Roudometof, 2016), namely, the constitution of sameness-cum-variation across organisations due to their exposure to global and local influences (Drori, 2016; Drori, Höllerer, & Walgenbach, 2014b, 2014a).

With this study, we contribute to world society theory by showing that organisational actors are not uniformly formed by globalisation. We employ a glocalisation perspective and demonstrate the existence of sameness-cum-variation across organisations in a supralocal field and three local fields in world society (Drori, 2016; Höllerer, Walgenbach, & Drori, 2017; Walgenbach, Drori, & Höllerer, 2017). Our study points to the significance of the glocal orientation, which organisations develop by coalescing global and local institutional demands. We argue that this glocal orientation is key to understanding the fine-grained and nuanced outcomes of globalisation processes. In detail, our findings indicate that organisations consider global and local institutional demands simultaneously and that it is the degree of orientation toward the global—or away from it—that characterises a focal organisation (Drori, 2016). In this way, a focal organisation’s glocal orientation involves sameness-cum-variation with regard to other organisations in the same field.

According to the glocalisation perspective, the glocal orientation of organisations manifests on three interconnected dimensions, namely, the formal, practice, and meaning dimensions (Höllerer et al., 2017). In our study we concentrate on the formal and meaning dimensions because our focus is on adoption, not on practice enactment. The formal dimension reveals the extent of the adoption of globally diffusing formal templates (such as the Global Reporting Initiative [GRI], ISO 14001) by organisations. As structural elements, formal templates build on global ideas (Meyer & Rowan, 1977). We do not consider the actual enactment of these templates (that is, the practice dimension) because, in the context of organisations’ orientation toward the global, the meaning dimension is prioritised (Höllerer et al., 2017). The meaning dimension captures not only formal conformity but also the degree of an organisation’s identification with the global. In this sense, on their meaning dimension, organisations display in which way they adopted and interpreted globally diffusing ideas such as responsible actorhood (Höllerer et al., 2017; Walgenbach et al., 2017).

In the following, we present our research context, formulate research questions, and explicate how our answers to these questions contribute to world society studies on the glocalisation of organisations (Drori, Höllerer, & Walgenbach, 2014a).

Empirically, we investigate the existence of sameness-cum-variation in the orientation of organisations as displayed in their self-representations concerning their responsibilities. These self-representations reveal the meaning that an organisation publicly attributes to the idea of responsibility. We focus on responsibility because the Western, now global, culture of modernity constructs organisations as sovereign actors and constitutes responsibility as the central element of modern organisational actorhood and agency (Drori et al., 2009; Hironaka, 2014; Meyer & Jepperson, 2000; Meyer, Pope, & Isaacson, 2015). The adoption and representation of responsible agency thus purge organisations in general and corporations in particular of the stigma of amorality (Boli, 2006).

We research publicly listed corporations from three economically powerful nation-states that—by belonging to the core of the Western world—can be regarded as deeply embedded in the cultural web of world society, namely, Germany, the United Kingdom, and the United States. We consider the analysis of the responsibility of corporations in these countries to be fruitful because patterns of sameness-cum-variation in these contexts should reveal whether the global idea of responsibility (Boli, 2006; Drori, 2008)—even in the case of countries that are intensively steeped in the supralocal sphere of world society—remains linked to historical and local institutional traditions (Brammer, Jackson, & Matten, 2012).

With our first research question, we disentangle the formal and meaning dimensions of organisations. We ask the following: Does the adoption of global formal templates concerning responsible actorhood correspond with the orientation of organisations regarding their responsibilities, as displayed on their meaning dimension? Previous research in the context of responsible actorhood has focused on the adoption of singular formal templates (for an overview, see Pope & Meyer, 2016) and, consequently, similarity on the formal dimension. The remaining variation across organisations was considered to be the result of diverging traditions and conditions on the local and organisational levels (Berliner & Prakash, 2013; Chen & Bouvain, 2009; Delmas & Montes-Sancho, 2011; Lim, 2017; Lim & Tsutsui, 2012; Vidal, Kozak, & Hansen, 2015). This focus appears problematic because—as Höllerer et al. (2017) recently argued—rather than factual conditions, it is the degree of orientation toward the global or local on the meaning dimension that accounts for sameness-cum-variation of organisations. In which regard the formal and meaning dimensions are co-constituted through glocalisation, however, remains an open question. Additionally, the focus on the formal dimension in studies on the adoption of singular formal templates neglects the potential effect of multiple diffusion (Pope, 2015; Pope & Meyer, 2016). Multiple diffusion and the adoption of multiple formal templates referring to different but mutually supporting facets of one and the same global idea may be accompanied by similarity on the meaning dimension of organisations.

With our second research question, we focus more specifically on the meaning dimension and take the local field into account. We ask the following: Does the orientation that organisations display on their meaning dimension vary with their local origin? With this question, we investigate whether the diffusion of global ideas is conditioned by the institutional traditions of the local field (contingent diffusion in the terms of Pope & Meyer, 2016) and thus implicates multiple glocalised forms of these global ideas (Roudometof, 2016). To date, researchers have focused on the contingent diffusion of formal templates into local fields (Boyle, Kim, & Longhofer, 2015; Lim, 2017; Lim & Tsutsui, 2012). However, researchers have not considered whether and how organisations interpret and display—depending on the institutional traditions of the local field from which they originate—global ideas on their meaning dimension (Drori et al., 2014a, 2014b). Therefore, addressing the orientation of organisations on the meaning dimension is relevant to better understanding the result of the simultaneous exposure of organisations to both global and local ideas about what denotes responsible actorhood (Brammer et al., 2012; Campbell, 2007; Pope & Meyer, 2016; Shanahan & Khagram, 2006).

With our third research question, we again focus on the meaning dimension. We ask the following: Do the orientations on the meaning dimension of organisations that originate from similar local fields correspond with the specific and historically evolved institutional traditions of these local fields? Addressing this question is relevant to empirically elaborate on whether the institutional traditions of similar local fields evoke similar patterns of sameness-cum-variation (Drori, 2016; Drori et al., 2014a). To date, existing researchers have suggested that the way in which formal templates are adopted is conditioned by the institutional traditions of a specific local field (Gjølberg, 2009; Lim & Tsutsui, 2012; Matten & Moon, 2008; Shanahan & Khagram, 2006). With our study, we thus contribute to this literature by investigating whether similar traditions in the local origins of organisations affect the gestalt of their orientation on the meaning dimension in a similar way.

Our paper is structured as follows. First, we introduce our key arguments and derive our hypotheses. We then present our methodological approach and the findings of our study. Finally, we elaborate on the implications of our study for the world society and glocalisation perspectives in institutional theory and then suggest further research.

According to world society theory, modern culture builds on the core idea that everyday life is structured in standardised means-end relations that serve progress and justice (Bromley & Meyer, 2015; Meyer & Bromley, 2013; Meyer, Boli, & Thomas, 1994; Meyer, Drori, & Hwang, 2006). This becomes visible, for example, in increasing consideration of sustainability (Weber & Soderstrom, 2015) and the protection of the natural environment (Frank, Hironaka, & Schofer, 2000; Hironaka, 2014), the enforcement of human rights (Hafner-Burton & Tsutsui, 2005), or the expansion of higher education (Schofer & Meyer, 2005). In this context, organisations are considered to be culturally constructed sovereign actors that, as rationalised agents, assume the responsibility to promote these global ideas in world society. World society also provides the formal templates (for example, ISO 14001) of what legitimate actorhood should look like (Bromley & Meyer, 2017; Drori et al., 2009).

Meyer and Jepperson (2000) argued that organisational actorhood implies assuming responsibilities in terms of four different but interrelated types of responsible agency (Meyer & Jepperson, 2000), namely, (1) agency on behalf of the self, (2) agency on behalf of other actors, (3) agency on behalf of units that are not construed as actors, and (4) agency in the service of abstract but highly esteemed cultural principles. Responsibility is thus considered the ultimate principle for organisations to act in a legitimate way in modern societies (Bondy, Moon, & Matten, 2012; Meyer et al., 2015; Shanahan & Khagram, 2006). By adopting the global idea of responsible actorhood, organisations should, subsequently, display responsibilities associated with the four types of responsible agency on their meaning dimension. However, the specific responsibilities of organisations are inherently cultural constructions that cannot be treated as ontologically given and definable from the outset (Campbell, 2007; Reinecke & Ansari, 2016).

World society theory argues that organisations are cognisant of the institutionalised global idea of responsible actorhood (Meyer et al., 2015). However, organisations are necessarily rooted in both supralocal and local fields. Therefore, they are simultaneously exposed to global ideas and (potentially conflicting) local ideas that reflect the nature of the traditional responsibilities in a local field (Brammer et al., 2012; Gjølberg, 2009; Pope & Meyer, 2016; Shanahan & Khagram, 2006). The simultaneous exposure of organisations to both the global and local induces what is described as ‘glocalisation’ or ‘glocality’ (Robertson, 1995; Roudometof, 2016). According to the glocalisation perspective, ‘the global is an abstraction and aggregate of various local instantiations of the very same abstract idea, floating in the global sphere and being repeatedly enacted in local contexts’ (Höllerer et al., 2017, p. 221). In other words, these various local instantiations are actually glocalised forms of responsibilities that result from coalescing global and local demands (cf. Roudometof, 2016).

Sameness-cum-variation across organisations indicates that the way in which glocalised forms of responsibilities are created necessarily depends on the degree of the orientation of an organisation toward the global or away from it (Walgenbach et al., 2017). Responsible actorhood on the formal dimension is exposed by the extent of the adoption of globally diffusing formal templates (Berliner & Prakash, 2013; Boli, 2006; Chen & Bouvain, 2009; Lim & Tsutsui, 2012; Pope, 2015; Shanahan & Khagram, 2006), which tends to increase similarity on the formal dimension across organisations in different local fields (DiMaggio & Powell, 1983; Drori et al., 2014a). These templates include the adoption of environmental (Berliner & Prakash, 2013) or accounting standards (Chen & Bouvain, 2009; Lim, 2017), membership in frameworks such as the United Nations Global Compact (UNGC; Lim & Tsutsui, 2012; Pope, 2015), or the adoption of specific organisational structures promoting responsible behaviour (Bondy et al., 2012; Matten & Moon, 2008).

The meaning dimension, which captures not only conformity with but also the degree of an organisation’s identification with the global (Höllerer et al., 2017), echoes the cognitive pillar of institutions (Scott, 2014) and thus reveals the cognitive orientation of an organisation. The cognitive orientation results from sense-making processes within an organisation and, finally, is mirrored in the self-representation that an organisation displays (cf. Höllerer et al., 2017). In the case of responsibility, the meaning dimension reveals how an organisation is cognitively oriented toward the idea of responsibility. For instance, an organisation with a high degree of cognitive orientation toward the global is likely to display a high identification with the idea of responsibility regardless of its local origin (cf. Höllerer et al., 2017) and to largely disregard the ideas of responsibility of its local field (cf. Brammer et al., 2012).

Formal templates and cognitive orientation. Previous studies on responsible actorhood have focused on similarity on the formal dimension (see Pope & Meyer, 2016). In these studies, variation across organisations was primarily traced back to diverging features of the local institutional and/or organisational level. On the local institutional level, variation was related to the propensity of political authority to supervise organisations (Lim & Tsutsui, 2012), the existence of specific communicative frames in local discourses (Boyle et al., 2015), or the existence of regulations regarding responsibility (Gjølberg, 2009; Lim, 2017). On the organisational level, differences were seen as grounded in the factual conditions of the organisation, for example, size and industry (Berliner & Prakash, 2013; Chen & Bouvain, 2009; Delmas & Montes-Sancho, 2011; Vidal et al., 2015).

However, as recently argued, besides factual conditions, it is the orientation of organisations—most importantly, the cognitive orientation of organisations (Höllerer et al., 2017)—that accounts for sameness-cum-variation. In which regard the formal and meaning dimensions are co-constituted through glocalisation remains an open question. Further, the dominant focus on the adoption of singular formal templates in previous studies neglected the potential effect that the adoption of multiple formal templates by an organisation may have on its cognitive orientation.

Indeed, the global idea of responsibility is the source of a multitude of formal templates that spread globally, that is, across local fields (multiple diffusion; Pope & Meyer, 2016). These templates address different facets of the very same global idea (Boli, 2006; Meyer et al., 2015; Shanahan & Khagram, 2006) and, taken together, represent the idea’s current state (Frank et al., 2000; Hironaka, 2014; Meyer, Frank, Hironaka, Schofer, & Tuma, 1997; Pope & Meyer, 2016). However, it remains unclear whether the adoption of multiple formal templates of the same global idea is also associated with a more comprehensive cognitive orientation on the meaning dimension of organisations toward this idea. Thus, our understanding of the relationship of these two aspects of organisations’ similarity – that is, the association of the formal and meaning dimensions – is limited. It is unclear whether similarity with regard to the formal dimension is associated with similarity in the cognitive orientation of organisations in supralocal fields in world society. It is at least the case, however, that several authors have argued that the adoption of formal templates may correspond with a specific cognitive orientation (cf. Fiss & Zajac, 2006; Haack, Schoeneborn, & Wickert, 2012; MacLean & Behnam, 2010; Tilcsik, 2010).

Considering that multiple formal templates may represent different facets of responsibility, which in their entirety represent the global idea (Pope & Meyer, 2016), the adoption of multiple formal templates by an organisation should thus be associated with the adoption of a more comprehensive cognitive orientation toward this global idea. This cognitive orientation should, regardless of the local origin of organisations, display responsibility more comprehensively and should be accompanied by similarity on the meaning dimension across these organisations.

Cognitive orientation and local origin. The sameness-cum-variation argument (Drori et al., 2014b, 2014a) highlights that increasing formal isomorphism in world society does not imply homogeneity across organisations (Drori, 2016). Instead, sameness-cum-variation points to the relevance of the cognitive orientation of organisations, that is, the identification with the global or the local (Höllerer et al., 2017; Walgenbach et al., 2017). This perspective thus offers a fruitful way to account for variations across organisations from different local fields (Drori et al., 2014b, 2014a). In particular, the adoption of the global idea of responsibility on the meaning dimension and the adoption of formal templates referring to this idea may be conditioned by the institutional traditions of the respective local field from which an organisation originates (Brammer et al., 2012; Campbell, 2007; Gjølberg, 2009). Indeed, although formal templates and ideas diffuse across local environments, the institutional traditions of a local field may influence their adoption (Malets & Quack, 2014; Pope & Meyer, 2016; Strang, 2014). Subsequently, global and local institutional demands coalesce into multiple glocalised forms of responsibility (cf. Roudometof, 2016). To date, however, researchers have focused on the contingent diffusion of global formal templates in local fields (Boyle et al., 2015; Lim & Tsutsui, 2012); it has not considered that the local origin might also evoke variation in the cognitive orientation of organisations. We focus on this gap.

Cognitive orientation and different local traditions of responsibility. We argue that specific local institutional traditions influence how contingent diffusion affects the cognitive orientation of organisations. Indeed, past research has suggested that the way in which formal templates diffuse is conditioned by the historically formed receptivity of the local field (Frank et al., 2000; Gjølberg, 2009; Lim, 2017; Matten & Moon, 2008; Meyer et al., 2015; Shanahan & Khagram, 2006). However, it neglected empirically investigating whether and how institutional traditions of the local field from which an organisation originates affect its cognitive orientation and thus the way in which it displays global ideas.

Referring to the responsibility of organisations, past research has distinguished between two kinds of institutional traditions that are linked to specific local fields. Matten and Moon (2008) proposed that organisations that have their origin in Europe are subject to an implicit variant of responsibility and therefore differ significantly in their orientations toward responsibility from organisations located in fields that hold explicit conceptions of responsibility. These authors argued that the implicit variant of responsibility forces organisations to adapt to a local consensus. Matten and Moon (2008, p. 410) stated that in these fields, ‘individual organizations would not normally articulate their own versions of such responsibilities.’ The orientation toward local consensus builds on ‘values, norms, and rules that result in (mandatory and customary) requirements for organisations to address stakeholder issues and that define proper obligations of corporate actors in collective rather than individual terms’ (Matten & Moon, 2008, p. 409). The implication is that organisations originating from fields with an implicit understanding of responsibility should exhibit a lower degree of variation in the cognitive orientation that they display. In contrast, organisations located in fields affiliated with the explicit variant of responsibility tend to expose their own orientation regarding their responsibilities (Matten & Moon, 2008), which is rooted in individual cost–benefit analyses (Flammer, 2015; McWilliams & Siegel, 2001; Porter & Kramer, 2006).

Matten and Moon (2008) emphasised that even if local fields affiliated with the implicit variant of responsibility have moved toward the explicit variant in recent years, they are still more closely related to local traditions that support responsible action based on the local consensus. We follow their argument. We expect organisations residing in fields affiliated with the implicit variant of responsibility to be more similar to each other, that is, in the way in which they display the local consensus of their responsibilities. Referring to Matten and Moon (2008), we also propose that organisations in fields that impose the explicit variant of responsibility tend to articulate their own versions of responsibility. As a result, these organisations should differ more from one another in terms of their cognitive orientation regarding responsibility than organisations that originate from local fields with an implicit tradition of responsibility. Therefore, the following hypothesis should hold:

We analysed the English-language websites of the largest publicly listed corporations in Germany, the United Kingdom, and the United States. According to the glocalisation perspective, the ‘global’ and the ‘local’ are relative categories, that is, the meaning of one category is dependent on the meaning given to the other (Drori et al., 2014a; Roudometof, 2016). We operationalised the local fields as national fields and the supralocal sphere in world society as the supranational field of corporations that embraces and steeps the three different national fields in our study.

Together, the corporations that we considered represent a large part of the global economy. In 2014, their combined sales comprised 7% of the world gross domestic product (WGDP), which in that year was $77.3 trillion USD. We chose these three countries because, together, their GDP represented approximately 30% of the WGDP in 2014. As Western countries are assumed to be deeply embedded in world society, we argue that corporations from these national fields should already exhibit a certain orientation toward the global, which we operationalised as the supranational orientation exhibited by the corporations of our sample. We argue that if the corporations in these countries show the hypothesised effects (see H2, H3), then we can conclude that national institutional traditions still have an important impact.

Jepperson (2002) highlighted that Western nation states differ on two dimensions, namely, in their degree of corporateness and centralism. These two dimensions correspond with the institutional traditions regarding the responsibility of corporations in national fields. According to Matten and Moon (2008), the United States, a liberal country, is oriented toward the explicit variant of responsibility and perceived as the main driver behind the global spread of the explicit variant of responsibility. The United Kingdom – which is characterised by a considerable orientation toward the implicit variant of responsibility, secured by state-enforced regulations (cf. Jepperson, 2002) – can be ascribed an intermediate position between the United States and continental Europe. Germany is considered to be a representative of the continental European tradition of rather centralised nation-states (cf. Jepperson, 2002), which is closely associated with the implicit variant of responsibility (Brammer et al., 2012; Matten & Moon, 2008).

We collected English-language web-based self-representations (for similar approaches using websites, see Park, Lee, & Hong, 2016; Powell, Horvath, & Brandtner, 2016) of the largest publicly listed corporations in Germany, the United Kingdom, and the United States, that is, from the German stock index (DAX), the British Financial Times Stock Exchange (FTSE), and the Dow Jones Industrial Average (DJIA) indexes, respectively. As the DAX and DJIA both contain the 30 most intensively traded and most highly valued corporations in their respective countries, we selected the 30 most intensively traded and most highly valued corporations from the FTSE 100, thus enabling a cross-index comparison. As the website of The Home Depot, one of the DJIA corporations, containing the necessary corporate information was not accessible, we excluded this corporation from our sample, which reduced our N to 89 corporations.

In general, we consider the selected corporations to be suitable to test our hypotheses because, first, the international website of the corporations is published by the corporate headquarters, and, second, all corporations have their main headquarters in the country from which they originate. Only three UK corporations have additional headquarters in foreign countries, although in neither Germany nor the United States. Third, we consider these corporations to be comparable because—with the exception of one US commodity and utility provider—they are all multinationals, that is, they all have subsidiaries abroad. Further, they are all listed on multiple and generally the same international stock indexes around the world (four, on average), whereby they are always listed in the country of their main headquarters. Fourth, the corporations in the three countries almost evenly span six industries: (1) information technology and services; (2) manufacturing; (3) chemistry, pharmacy, and health; (4) consumer products; (5) commodities and utilities; and (6) financial services.

For each corporation, we extracted all communication about its responsibilities from its corporate website using a self-adjusted version of the Nutch¹ web crawler, with which we crawled all of the websites between 10 April 2015, and 30 June 2015. Starting from the front page of every corporation’s website, all HTML and PDF documents were searched for our node words: ‘responsibility’ and ‘responsibilities.’² If the node words were found, then the documents were downloaded and subsequently decomposed into separate paragraphs. This process also included headings and captions. To increase comparability, we standardised the paragraphs manually by adjusting any divergent spellings in British English to American English. Furthermore, we checked whether the native-language websites of the German corporations encompassed more or divergent content. We found that the German websites were translated into English with a one-to-one correspondence. Finally, an initial review of the data revealed that many passages occurred repeatedly on different websites of the same corporation, which means that the same paragraph appeared in the data several times. After automatically removing identical³ paragraphs, we identified 84,623 unique paragraphs containing 44,712,784 words, which we used for all remaining analyses.

We operationalised the cognitive orientations of the corporations in our sample through the vocabulary structures that they use to express their responsibilities, that is, in our case, word-to-word relations (Loewenstein, Ocasio, & Jones, 2012). We focused on relevant word-to-word relationships, namely, on specific word classes that are grammatically associated with ‘responsibility’ and ‘responsibilities’ in comparable social and linguistic contexts (Hoey, 2005; Kilgarriff, 1997; Sinclair, 1991). These so-called grammar patterns are defined as linguistic expressions comprising grammatical relations (for example, the grammatically meaningful connection of ‘responsibility’ to a verb and an object) that reveal how words are regularly embedded in communication (Hunston & Francis, 1999). In this respect, grammatical relations focus on the most objective level of language and explicitly exclude linguistic artifacts, such as personal style or country-specific language norms.

We selected grammar patterns that are relevant according to our theoretical considerations. From our theoretical perspective, organisations have the obligation to assume responsibility for the self, other actors, non-actors, and cultural principles. Organisations can linguistically display these responsibilities by referring to verbs (what kind of actions are associated with responsibility?), adjectives (what kind of responsibility is assumed?), and subjects and objects (who takes responsibility on behalf of whom and for what?). In summary, we defined the lemma⁴ (‘responsibility’) as the node word within the grammar patterns representing the four types of responsible agency (see Table 1).

Grammar patterns were the relevant linguistic context for our analysis. The bolded elements in Table 1 below are considered destination words because they occupy the particular grammatical positions (i.e. subject, adjective, object, or verb) in which we are interested (Table 1 provides examples of destination words). All destination words are, in turn, lemmatised to increase comparability.

We used computational linguistics, especially natural language processing (NLP), to extract the destination words from the grammar patterns (for an overview, see Evans & Aceves, 2016). We applied dependency parsing to analyse our data (for a recent call to do so in organisation studies, see Cornelissen & Werner, 2014). Dependency parsing is well suited for an automatic analysis of the grammar of sentences (for a comparable approach applied in the social sciences, see Sudhahar, de Fazio, Franzosi, & Cristianini, 2015; van Atteveldt, Kleinnijenhuis, & Ruigrok, 2008). The parser detects the grammar patterns within each sentence, within which each word relates to at least one other word in a form of linguistic dependency (Carroll, 2004; Klein & Manning, 2003). The parser then derives from these linguistic links the grammatical functions of words (such as subjects, adjectives, objects, and verbs linguistically related to responsibility) and thus the underlying grammar patterns as a whole (de Marneffe, MacCartney, & Manning, 2006). We used the Stanford Parser⁵ (Cer, de Marneffe, Jurafsky, & Manning, 2010; Klein & Manning, 2003).

Dependent variables. In our study, we used two different dependent variables. Both variables represent vocabulary structures to measure the variation between the cognitive orientation that an individual corporation displays and the cognitive orientations in the supranational field (all corporations in our sample) or in a national field (all corporations with the same national origin in our sample). We found this approach to be suitable because world society theory considers the global to be an aggregate of various local instantiations of the same idea (Höllerer et al., 2017). The implication is that each corporation communicatively displays its instantiation of responsibility. A national field of corporations, in turn, forms the local instantiation of responsibility because it encompasses all the destination words to which members of this field refer. Finally, the aggregate of the destination words that corporations of a supranational field use to display their responsibilities encompasses these various local instantiations. It is important to note that we do not claim to measure the global in its most abstract sense but, instead, the aggregate of instantiations across comparable entities in three national fields.

Our first variable measured the variation between the cognitive orientations displayed on corporate websites and the cognitive orientations of all corporations in our sample. We used this variable to test H1 and H2. Our second variable measured the variation between the cognitive orientations displayed on corporate websites and the cognitive orientations within a national field. We used this variable to test H3. For both dependent variables, we calculated the distance between the cognitive orientations regarding responsibility that an individual corporation displays and the two aggregates of cognitive orientations.

To calculate these distances, we first needed to extract the cognitive orientations from the bundle of destination words that we collected. Referring to Kilgarriff (1997), we reconstructed the cognitive orientation by clusters of word-to-word relations in a specific linguistic context. These linguistic contexts were, as outlined above, the grammar patterns of the lemma ‘responsibility’ that cluster on the websites of the corporations in our sample. Therefore, each corporation holds a collection of destination words that together comprise the corporation’s representation of its responsibilities. The implication is that if corporations use similar destination words with a similar frequency on their websites, then they also exhibit similar cognitive orientations regarding their responsibilities.

We quantified and compared the cognitive orientations by representing them through vectors. We added the destination words used by a specific corporation on its website as the dimensions of a vector space (Schütze, 1998; Turney & Pantel, 2010). In this regard, we followed recent organisational and social-science research that attributed a high relevance to vector spaces for research (Ellsaesser, Tsang, & Runde, 2014; Evans & Aceves, 2016) and that already applied vector spaces in empirical studies (Piezunka & Dahlander, 2015). Vector spaces are especially well suited to uncover the semantic distance between the textual embedding of words (such as responsibility) in different sources (such as websites of corporations). The basics of semantic vector spaces are related to other techniques that were already successfully applied in social science research (such as topic modeling; Bonilla & Grimmer, 2013; DiMaggio, Nag, & Blei, 2013; Fligstein, Brundage, & Schultz, 2017; Mohr & Bogdanov, 2013). In our case, the value of each of the dimensions within the vectors was represented by the count of a particular destination word, which was normalised by the sum of the counts of all destination words on the respective website. Thus, a vector captures the cognitive orientation that an individual corporation displays through all destination words that are grammatically related to responsibility (cf. Padó & Lapata, 2007).

Independent variables.Formal templates concerning responsibility. Corporations may deploy a range of templates to demonstrate their commitment to responsible actorhood. Following Waddock (2008), we divided these formal templates into subgroups related to (1) general principles, (2) reporting, (3) certifications, (4) rankings, and (5) the use of consulting agencies. We added another group (6) that encompasses the organisational integration of responsibility (for example, job roles or organisational units). For each category, we referred to one formal template that is prominently discussed in the literature (Boli, 2006; Shanahan & Khagram, 2006; Waddock, 2008) and that the corporations in our sample might use.

(1) An important collection of principles concerning responsibility are included in the United Nations Global Compact (Shanahan & Khagram, 2006; Waddock, 2008). Researchers have shown that membership in the UNGC increases the probability that an organisation will adopt other templates or address specific topics related to responsible corporate behaviour (Chen & Bouvain, 2009). (2) Among the numerous reporting standards available, we choose the Global Reporting Initiative, as it is the pioneering initiative of the accounting movement concerning responsible actorhood (Shanahan & Khagram, 2006). (3) ISO 14001 represents a widespread environmental standard that is accredited by external auditors (Berliner & Prakash, 2013; Boli, 2006). This external accreditation clearly distinguishes the UNGC and ISO 14001 from other standards, such as ISO 26000, the application of which is neither accredited nor tied to an organisational network. We argue that the accreditation process is likely to support the adoption of a specific cognitive orientation regarding responsibility. (4) Indexes measuring the responsible behaviours of corporations, such as the Dow Jones Sustainability Index (DJSI), were launched to meet the expectation that responsible corporate behaviour will lead to superior economic performance (López, Garcia, & Rodriguez, 2007). (5) Waddock (2008, p. 93) highlighted that the growing number of formal templates concerning responsibility has led to a burgeoning of consultancies (for example, SalterBaxter MSL Group, DNV GL) that develop ‘practices geared toward helping companies improve their responsibility performance in one way or another.’ In line with this argument, we argue that consultants convey a specific cognitive orientation regarding responsibility. (6) An important sign of the incorporation of responsible actorhood is the adoption of structural elements, such as specific jobs or departments (cf. Fiss & Zajac, 2006) whose titles are formalised in organisation charts or in job descriptions (such as Chief Sustainability Officer, Corporate Responsibility Manager).

Data for these measures were extracted from the corporate websites, which were cross-checked with external sources (such as the websites of the UNGC or consultancies). The variable regarding consulting agencies was coded as a dummy variable, whereas the remaining formal responsibility templates were summed for each corporation to measure the extent of the adoption of formal templates. We opted for these measures because the adoption of a formal template is different from making use of a consultant agency (Boli, 2006; Waddock, 2008).

Countries. To measure the effect of national origin, we used three dummy variables for the corporations originating in Germany, the United Kingdom, or the United States.

Control variables. We used control variables that were already applied in earlier studies as explanatory variables for variation among organisations.

Size. The size of an organisation may affect how responsibility is displayed, as size is an indicator of the degree of institutional pressure that an organisation faces (Dobbin, Edelman, Meyer, Scott, & Swidler, 1988). We used the total volume of sales in 2014 as a measure for size (Stanwick & Stanwick, 1998; Waddock & Graves, 1997) and calculated a z-score value.

Ownership. The representation of responsibility might also depend on the distribution of shares. We control for the free float in 2014 and standardise all values into z-scores. We also control for the power of majority shareholders. We use three categories: No shareholder holds more than 25% of all shares, one shareholder owns between 25% and 50% of all stock, and one shareholder holds more than 50% of all shares. We assigned values from 1 to 3 to indicate increasing shareholder control. Information for European corporations was extracted from the Amadeus database of Bureau van Dijk⁶ and from corporate websites. The data for American corporations were extracted from the EDGAR database (US Securities and Exchange Commission)⁷ and from corporate websites.

Age. Matten and Moon (2008) argued that the implicit variant of responsibility is gradually being replaced by the explicit variant. However, organisations in Europe – especially older organisations – may be particularly committed to the implicit variant of responsibility. We therefore control for the age of the corporations to ensure that deviations from the cognitive orientation in the supranational and national fields may not be traced back to organisational inertia (Hannan & Freeman, 1977).

Financial performance. The way in which corporations understand their responsibilities may also depend on their financial performance (Orlitzky, Schmidt, & Rynes, 2003). We measure financial performance using an accounting-based variable and a market-based variable, namely, total returns, and total return on assets (ROA) in 2014 (Fiss & Zajac, 2006), respectively. Data were extracted from the Amadeus and EDGAR databases and from corporate websites and then standardised into z-scores.

Internationalisation. Internationalisation is measured by the number of countries in which a corporation operates (cf. Walgenbach et al., 2017). We also control for the number of international subsidiaries. Data were extracted from the Amadeus and EDGAR databases and from corporate websites and then standardised into z-scores.

Visibility. The visibility of an organisation is considered as an indicator of the institutional pressure that it faces (Dobbin et al., 1988). Institutional pressure may affect how a corporation displays its responsibilities. We measured visibility by counting the number of newspaper articles that mentioned a corporation in 2014 (Meznar & Nigh, 1995). For Germany, we used Handelsblatt, a frequently read daily publication in economics and business, and the Frankfurter Allgemeine Zeitung, one of the most widely read daily newspapers. For the United Kingdom, we selected The Guardian and The Independent, which are both widely circulated. For the United States, we collected data from The New York Times, a publication read by a large proportion of the American population, and The Washington Post, a nationally distributed newspaper with a focus on Washington, DC. We calculated the average of the scores of all newspapers and standardised the averages into z-scores.⁸

Industry. It has often been argued that firms operating in the same industry become increasingly alike (Leblebici, Salancik, Copay, & King, 1991). Firms in the same industry accomplish comparable tasks, which may lead to greater similarity in their reactions to institutional demands. To control for the industry, we established six dummy variables according to the classifications of the Deutsche Börse Group: (1) information technology and services; (2) manufacturing; (3) chemistry, pharmacy, and health; (4) consumer products; (5) commodities and utilities; and (6) financial services. In our regression analyses, we used information technology and service as a reference category.

Table 2 presents general information on our textual data and the destination words used in our study. First, the paragraphs from each country are comparable with regard to their length (see rows 1 and 2). Second, Table 2 shows that the cognitive orientations regarding the responsibility of all corporations in our sample are not biased by unique destination words, that is, words that appear only in the respective countries (row 3). Additionally, as shown by the widely normally distributed contribution of unique destination words, the cognitive orientations in the national fields are not driven by only a few corporations within the respective countries (row 4).

Table 3 provides the descriptive statistics and Pearson’s correlation coefficients for the variables that we used to test H1 and H2. Table 4 shows the results for our ordinary least squares (OLS) regressions. We proceed stepwise (Models 1 to 6), starting with the control variables. Then, in the next step, we add our variables for formal templates and the national origins of the corporations. In all models, none of the control variables, except visibility (p < 0.10) and free float (only in the full model, Model 6), show a significant influence on the distance between the cognitive orientation of individual corporations and the cognitive orientations of all corporations in our sample. Visibility is significant in Models 3 to 5 but not in Models 1, 2, and 6. However, for Models 1, 2, and 6, the results are similar to those in Models 3 to 5, that is, in direction and effect size. They evade the significance level of p < 0.10 only very slightly. In Model 1, the corr. R² is low, with a value of 0.08, but it increases to 0.17 in Model 2. As the beta coefficients show, the extent of the adoption of formal templates (p < 0.05) and the use of consultancies (p < 0.10) are significantly associated with a smaller distance to the cognitive orientations in the supranational field. In a robustness check, we further substantiated these results by including all formal templates separately in Model 2. In this case, only the consultancies variable shows a significant effect (p < 0.10). This result implies that only the adoption of multiple formal templates, not the adoption of single formal templates, by a corporation shows the reported effect. In Models 3 to 5, we add the country variables, whereupon the model fit increases, corr. R² = 0.16. These results show the significant influence of the country variables (p < 0.01). Finally, Model 6 represents the full model, including formal templates, consultancies, and national origin. We present only the model with the United States as the reference category because, by using Germany and the United Kingdom as reference categories, the effects remain consistent with Models 3 to 5. The results show that the influence of the formal templates (p < 0.05) and the country variables is still significant (p < 0.05); however, the significant influence of consultancies disappears.

H1 suggested that the variation between the cognitive orientation of an organisation and the cognitive orientation displayed by organisations in a supralocal field in world society is smaller if the organisation has adopted a larger number of formal templates. Our results provide support for this hypothesis. H2 suggested that the variation between an organisation’s cognitive orientation and the cognitive orientations displayed by organisations in a supralocal field in world society corresponds with the local origin of the organisation. We find that in all countries, the country variable has a significant effect (p < 0.01 in Models 3 to 5 and p < 0.05 in Model 6), which supports H2. However, it is important to note that originating from the United States has a positive effect on the distance from the cognitive orientations in the supranational field, whereas originating from Germany or the United Kingdom has a negative effect on distance.

Table 5 provides the descriptive statistics and Pearson’s correlation coefficients for our tests of H3. Table 6 shows the results for our OLS regressions with regard to the aggregated cognitive orientations of the corporations in the three different national fields as the dependent variables and the country variables as the independent variables.

H3 suggested that organisations belonging to an environment associated with the implicit variant of responsibility are more likely to orient toward a national consensus regarding responsibility across organisations. Again, we proceed stepwise, starting with our control variables and adding our country variables. In Models 1 to 3, we use the distance from the national cognitive orientation in Germany. In Model 1, the corr. R² is low, with a value of 0.14, and only the amount of free float is significant (p < 0.10). However, after adding our independent variables in Models 2 and 3, this effect disappears, and the corr. R² rises to 0.46. These results indicate that, for Germany, only the country variable corresponds with the decrease in the distance from the cognitive orientations in the national field, with control variables having no impact on distance. This result stresses the low variation across German corporations. In Models 4 to 6, we use the distance from the cognitive orientations regarding responsibility in the United Kingdom. In Model 4, the corr. R² is low (0.13), whereas we again find a significant result of the influence of the free float (p < 0.10). For this country, the consumer products and commodities and utilities industries show significant effects (p < 0.05). Whereas the effects of the free float and of the commodities and utilities industry disappear in Models 5 and 6, the effect for the consumer products industry decreases to a significance level of p < 0.10. Visibility becomes significant in Models 5 and 6 (p < 0.05). Overall, the corr. R² increases to 0.57. The country variable (β = −0.72) mainly accounts for the similarity of corporations in the United Kingdom. In Models 7 to 9, we use the distance from the national cognitive orientations in the United States context. Again, the free float is significant (p < 0.10) in the model that includes only the control variables, with corr. R² = 0.13. Visibility is also significant at the p < 0.10 level across all models for the cognitive orientation in the United States.

Thus, with regard to visibility, the results are significant for the cognitive orientations in the British and American national fields (Models 4 to 9). The results for Germany are similar in direction and effect size, and they evade the significance level of p < 0.10 only very slightly (Models 1 to 3).

In the United States, however, we do not find any effect of the country variable, and the corr. R² decreases to 0.11. American corporations show a significantly higher degree of variation than their British or German counterparts. These results support our third hypothesis.

According to world society theory, organisations are culturally constructed as sovereign actors that adopt globally diffusing formal templates and ideas of organising. We contribute to this perspective by emphasising that the global is not to be considered the only force that influences organisations. It is rather the multiplicity of local and global institutional demands that organisations coalesce into a specific orientation toward the global and the local. We thus highlight that organisations’ glocal orientation is key to understand the fine-grained and nuanced outcomes of globalisation processes.

We empirically elaborated on the argument of sameness-cum-variation and investigated whether the adoption of multiple formal templates supporting responsible actorhood is associated with a similar cognitive orientation of organisations. Furthermore, we examined whether and how the local origin of organisations affects the variation in their cognitive orientation. We analysed the content of websites depicting cognitive orientations with regard to the issue of responsibility. In the following, we discuss our contributions to the literature on the world society and glocalisation perspectives in institutional theory (Drori, 2016; Drori et al., 2014a, 2014b; Höllerer et al., 2017).

Addressing our first research question, we studied whether the adoption of global formal templates corresponds with the cognitive orientation of organisations. Our study sheds light on the co-constitutive nature of the formal and meaning dimensions of organisations’ orientation (Höllerer et al., 2017; Walgenbach et al., 2017). In addition, we show that the adoption of multiple formal templates and the hiring of consultants not only function as ceremonial markers of sameness (Drori, 2016) but are also positively correlated with a supralocal orientation.

In particular, our results indicate that the adoption of multiple formal templates is associated with the degree of sameness between the cognitive orientations of organisations in the supralocal field. It is important to note that none of the formal templates alone has a significant effect, as indicated by our robustness check. One possible explanation for this finding is that the actual interpretation of single formal templates, which focus on one facet of the global idea of responsibility, may remain ambiguous (Czarniawska & Joerges, 1996; Guthey & Morsing, 2014; Strang & Meyer, 1993). However, our results provide evidence that—in correspondence with the idea of multiple diffusion (Pope & Meyer, 2016)—the adoption of multiple formal templates is indeed associated with a more similar representation of an idea, which means that organisations display a supralocal rather than a local cognitive orientation.

Additionally, our results suggest that the hiring of consultants has an impact on the similarity across organisations’ cognitive orientation in a supralocal field. Because the notion of responsibility is an abstract global idea (Boli, 2006), it tends to diffuse among organisational actors with the assistance of ‘culturally legitimated theorists’ (Strang & Meyer, 1993, p. 494). Management consultancies, although closely interrelated with other carriers of theorised and standardised knowledge (such as academia and the media), were denoted in past research as important culturally legitimised theorists (Sahlin-Andersson & Engwall, 2002). One salient attribute that distinguishes management consultancies from other carriers of knowledge is their direct interaction with clients. In this context, an argument established by Strang and Soule (1998) already implied that diffusion processes not only refer to the formal and practice dimensions but also occur on the meaning dimension. This aspect leads back to our results and to the argument that management consultancies not only carry theorised practices to organisations but also tend to influence the cognitive orientations of organisations.

With our second research question, we aimed to explore whether the orientation that organisations display on their meaning dimension varies with their local origin. Our results highlight the importance of contingent diffusion (Pope & Meyer, 2016) in glocalisation research (Drori et al., 2014b, 2014a). Our findings suggest that the variation between the cognitive orientation in a supralocal field of organisations and the cognitive orientations displayed in the self-representations of individual organisations cannot be explained by conditions such as size, ownership structure, industry, or the degree of internationalisation. It is foremost an organisation’s degree of identification with its local field that explains the variation between organisations in a supralocal field and that highlights sameness-cum-variation as an inherent feature of organisations in world society. This finding challenges the assumption that the aforementioned conditions are important parameters in the adoption of formal templates and global ideas (Walgenbach & Meyer, 2008). Our finding also corresponds with the notion that due to the standardised character of modern organisations, their meaning dimension and, consequently, their cognitive orientations are constructed regardless of the factual conditions of the organisation (Höllerer et al., 2017). Further, we stress the argument that the local field, even if steeped in the supralocal sphere of world society, significantly affects the cognitive orientations of organisations regardless of whether the respective organisations are large, operate in different industries, or follow globally oriented business strategies (cf. Drori, 2016; Höllerer et al., 2017; Tempel & Walgenbach, 2007).

Regarding our third research question, we provided evidence that the specific result of a contingent diffusion corresponds with the distinct institutional traditions of a specific local field. We found that in the local fields of German and British corporations, a lower degree of variation exists in the displayed orientations, whereas the opposite holds true for American organisations. This finding underlines Jepperson’s (2002) notion that Western nation states differ and gives support to Matten and Moon’s (2008) argument that a local consensus developed in environments that are affiliated more closely with the implicit variant of responsibility. In the case of Germany and Britain, these local consensuses appear to explain the short distances between the cognitive orientations in the local fields.

From a methodological perspective, our big data method is novel in social science research. Our approach may be expanded to all kinds of websites and thus to all kinds of organisations that have websites. Our data analysis, which draws on NLP (that is, dependency parsing and vector spaces), opens new avenues for quantitatively measuring vocabulary structures and meaning. Our paper shows how to turn grammatical structures into numerical values for analysis while simultaneously preserving the grammatical functions in the original texts. In this way, NLP provides a fruitful addition to the reservoir of linguistically and relationally oriented approaches in text analysis (Carley, 1997; Franzosi, 1989; Roberts, 1989).

Finally, our study’s limitations open avenues for future research. Although we investigated only large corporations in Germany, the United Kingdom, and the United States, we do not consider this sample to be a critical limitation given the scope and nature of our study. However, future research should consider more countries and/or a broader range of organisations.

A second limitation is that we did not consider the practice dimension, which would have shown how corporations actually enact their responsibilities (Höllerer et al., 2017). Focusing on the practice dimension could also clarify a conceptual ambiguity in the glocalisation perspective. While our findings showed a positive relationship between the orientation on the formal and meaning dimensions of organisations, Walgenbach et al. (2017) also pointed to the possibility of a negative relationship. In the latter case, a higher number of adopted formal templates should be associated with a lower degree of orientation toward the global and vice versa. The practice dimension might be a moderator of this relationship. Whether this assumption holds should be explored in future research. Studying the practice dimension in more depth, we believe, is also a fruitful way by which to investigate how the endemic phenomenon of decoupling in world society (Pope & Meyer, 2016) corresponds with the adoption of global templates and ideas on the formal and meaning dimensions of organisations. In addition, this line of research could reveal how the decoupling of both formal templates and practices as well as of practices and meaning (Li, 2017) is influenced by glocalisation.

A third limitation is that we did not consider the influence of time on the process of glocalisation. Focusing on time may reveal the reciprocal relationship between the formal and meaning dimensions of organisations. Further, a focus on time may uncover how the cognitive orientations regarding responsibility are constructed, as well as whether and how local understandings, for example understandings of responsibility, coalesce with understandings in world society.

We thank Michael Hunoldt, participants in the 13th New Institutionalism Workshop at Hebrew University and participants in Friedrich Schiller University’s research colloquium on management and organisation studies, and anonymous reviewers for their helpful comments.

No potential conflict of interest was reported by the authors.

Jan Goldensteinhttp://orcid.org/0000-0003-2538-0559

Philipp Poschmannhttp://orcid.org/0000-0002-7056-6013

Sebastian G.M. Händschkehttp://orcid.org/0000-0003-4888-7125