Implementing Socially Sustainable Practices in Challenging Institutional Contexts: Building Theory from Seven Developing Country Supplier Cases

Sustainability has been defined as a development that meets the needs of the present without compromising the ability of future generations to meet their own needs (Bruntland 1987). But this definition provides only a vague macro perspective that does not specify what resources are required by individual organizations to meet those needs (Carter and Rogers 2008). Elkington’s (1998) Triple Bottom Line (TBL) perspective on sustainability includes a broader range of issues, including economic, environmental, and social performance. From a global supply chain perspective, organizations operate within a network of inter-dependent relationships and it has been argued that understanding the economic, environmental, and social dimensions of sustainability is crucial (Vachon and Klassen 2008). There have thus been numerous conceptualizations of sustainability specifically in a supply chain context, including those by Carter and Rogers (2008) and Ahi and Searcy (2013). Meanwhile, Seuring and Müller (2008, p. 1700) defined sustainable supply chain management as “the management of material, information and capital flows as well as cooperation among companies along the supply chain while taking goals from all three dimensions of sustainable development, i.e., economic, environmental and social, into account which are derived from customer and stakeholder requirements.”

It is important to consider the relationships, synergies, and trade-offs amongst the economic, environmental, and social “pillars” of sustainability (Hutchins and Sutherland 2008; Haffar and Searcy 2017). It is however difficult to implement all three aspects simultaneously in practice and to capture them all in research in order to achieve “true” sustainability (Pagell and Shevchenko 2014). The trend in ethical supply chain management research has therefore been for studies to either focus on environmentally sustainable supply chain management (Ahi and Searcy 2013) or on socially sustainable supply chain management (Huq et al. 2014). It is clear from the numerous recent literature reviews that upstream social issues relating to the human/people-side of sustainability have been under-researched when compared to the environmental agenda (Yawar and Seuring 2015; Zorzini et al. 2015; Kim et al. 2016). Further, Touboulic and Walker (2015) noted that there has been more focus on correlations between environmental and economic performance than between social and economic performance because environmental performance is more quantifiable. Yet social sustainability requires further research in its own right; however, challenging it might be to study the phenomenon.

Social issues are different from environmental issues, and hence, it cannot be assumed that findings on the environmental dimension translate to the social dimension. Social issues are more relevant in certain sectors (e.g., labor intensive industries) and it is harder to measure social performance (Zorzini et al. 2015), mainly because of the dynamic and complex nature of social concerns (Yawar and Seuring 2015). Further, it is generally easier to agree what constitutes best environmental practice (e.g., carbon dioxide reduction) when compared to the social dimension (e.g., practices concerning the use of child labor) due to cultural disparities between countries (Huq et al. 2014; Jia et al. in press) that mean further context-specific research is required. Moreover, persistent social failures (e.g., the Rana Plaza collapse) affirm that there is insufficient knowledge of what drives organizations to engage in socially sustainable practices and what limits their involvement, especially the challenges of implementing such practices in developing country suppliers (Yawar and Seuring 2015). These suppliers are often situated in very different and challenging institutional contexts compared to their downstream supply chain partners yet, historically, investigation into the institutional environment and its role in implementation has been overlooked (Zhou et al. 2016).

To understand the motivations and impediments behind the implementation of socially sustainable practices in developing country suppliers, we employ institutional theory as our overarching theoretical anchor. Institutional theory provides a framework that explains how organizations gradually respond to a combination of pressures from actors within their institutional field by converging on a set of homogeneous business practices (DiMaggio and Powell 1983; Grewal and Dharwadkar 2002; Zsidisin et al. 2005), which become the legitimate way to organize (Meyer and Rowan 1977; Zucker 1987). The potential of institutional theory to improve understanding of operations and supply chain phenomena has been identified by a number of authors (Kauppi 2013). It has been used, for example, to unpack the drivers behind the adoption of business continuity planning procedures (Zsidisin et al. 2005), internet-enabled supply chain management systems (Liu et al. 2010), and inter-organizational systems across multiple tiers of a healthcare supply chain (Bhakoo and Choi 2013). But its use in the sustainability literature has been mainly limited to the study of environmental practices. To the best of our knowledge, institutional theory has not been used to study the implementation of socially sustainable practices (Touboulic and Walker 2015; Zorzini et al. 2015).

Institutional theory posits that firms adopt organizational practices mainly due to the following three types of pressure (DiMaggio and Powell 1983; Grewal and Dharwadkar 2002):

A wide range of institutional actors can exert pressures that influence what becomes legitimate (DiMaggio and Powell 1983; Eisenhardt 1988; Hoffman 2001; Scott 2008). The literature has identified buyer pressure as one of the most important factors behind the implementation of social standards (Maloni and Benton 2000; Luken and Stares 2005; Tencati et al. 2008; Yu 2008). Developing country suppliers face pressure from more powerful buyers who make socially sustainable practices a pre-condition to obtaining orders (Marshall et al. 2016). When an organization is highly dependent on the constituent exerting pressure, as in the case of Bangladeshi apparel suppliers and their Western buyers, it is considered unlikely that the institutional demands and expectations will be resisted (Benton and Maloni 2005; Zhao et al. 2008). Competition amongst suppliers can also be a major factor (Yu 2008; Park-Poaps and Rees 2010). This may help or hinder implementation, depending on whether social improvements are viewed as a cost or potential source of competitive advantage.

Internal actors, including owners/managers (Baden et al. 2009) and senior management (Walker and Jones 2012), are also influential while governments can impact social conduct through regulation and laws (Lim and Phillips 2008; Yu 2008; Lee and Kim 2009). It has also been suggested that NGOs (Mamic 2005; Tsoi 2010; Ehrgott et al. 2011) and trade unions (Lipschutz 2004) play a positive role in pressurizing firms into being socially sustainable. Finally, professional networks, including trade bodies, can have an effect, including by providing education on social sustainability issues (Hoffman 2001; Campbell 2007).

There are various tools for implementing socially sustainable practices in supply chains, ranging from a buyer’s own code of conduct to third-party standards and supplier development programmes. There remains however a lack of knowledge in the literature on how MNCs can implement socially sustainable practices across their supply chains to enhance social performance (Klassen and Vereecke 2012; Zorzini et al. 2015). Based on a thorough review of the supply chain ethics and sustainability literature, we identify six main composite practices. These six dimensions in particular were considered to be key to facilitating and encouraging social sustainability in developing country suppliers. Below we present a discussion on why these factors are important for our research context (see also Table 1, which includes a list of key references).

It is known that the general attitude of suppliers towards compliance is an important factor (Marshall et al. 2015), which can be reflected in the scope of the practices adopted. This scope can vary from compliance with local laws only to complete compliance (or beyond) with buyers’ own codes of conduct (Gugler and Shi 2009) or with third-party certifications (Ciliberti et al. 2009). Like the scope, the depth of the compliance behavior can also differ, with instances of suppliers carrying out ‘mock compliance’ being reported (Huq et al. 2014) and buyers subsequently displacing their responsibility (Eriksson and Svensson 2016). Infrastructure within supplier firms to support the implementation of socially sustainable practices also varies. It has been found that having dedicated staff or departments dealing with the implementation of socially sustainable practices leads to better implementation when compared with suppliers having no specialized staff, where a broad human resource department is left to oversee implementation (Gattiker and Carter 2010; Huq et al. 2016). The role of education and awareness in implementing socially sustainable practices across the firm has also been highlighted as a reason for better implementation (Mamic 2005; Awaysheh and Klassen 2010).

Meanwhile, an inter-organizational collaborative approach can help to develop social sustainability implementation capabilities (Klassen and Vereecke 2012; Eriksson and Svensson 2016). Firms can gain legitimacy by making stakeholders inclusive to the implementation process (Mueller et al. 2009) and the resources and capabilities that develop from such collaborations can lead to sustained competitive advantage, especially since these are difficult to imitate as they are the result of long-term, complex relationships (Gold et al. 2010).

Finally, recent research has shown that social sustainability leadership in terms of being proactive and striving for the continuous development of innovative implementation tools and procedures leads to better implementation (Huq et al. 2016). There is however limited knowledge on what drives implementation efforts in suppliers, especially in developing country suppliers (Ehrgott et al. 2011; Gimenez and Tachizawa 2012; Zhou et al. 2014).

The pressures exerted by institutional actors do not always lead to implementation. Instead, heterogeneous responses may occur, including compromise, avoidance, defiance, and manipulation (Oliver 1991). Meyer and Rowan (1977) suggested that an organization may respond to institutional pressures through superficial conformity. This “decoupling” of pressure from expected response arises when adaptations to institutional pressures have uncertain efficiency consequences (Meyer and Rowan 1977; Rogers et al. 2007; Boxenbaum and Jonsson 2008), if they contradict internal efficiency needs, i.e., create potential trade-offs in priorities (Meyer and Rowan 1977), when practices do not reflect local circumstances or realities (Scott 2008), or when there is information asymmetry between firms and their stakeholders (Crilly et al. 2012). A case in point is the apparel supply chain, which included a number of incidents of social failure indicating a decoupling between the socially sustainable policies adopted by buyers and the actual implementation of socially sustainable practices in suppliers. For example, several factory fires led up to the collapse of the Rana Plaza in Bangladesh in 2013, including a fire in a factory supplying apparel to major branded clothing companies such as Wal-Mart and SEARS in November 2012 (Tazreen), resulting in the loss of 117 lives (Bloomberg 2012). It was claimed that the Tazreen factory had been assessed by third-party auditors (New York Times 2012) yet post-fire investigations revealed the factory had no emergency exits and that the gates were locked from the outside (BBC 2014). It was also claimed workers were poorly trained on how to extinguish fires and on evacuation procedures. It seemed that, in many cases, the focal actors were complying superficially because being compliant legitimized their business. Suppliers were clearly decoupling formal compliance from ground-level organizational practices. Implementation may be viewed by suppliers as an extra cost, especially during constrained economic times (Barnett et al. 2015). Thus, if the economic benefit of implementation is not apparent then suppliers can act opportunistically (Huq et al. 2014) and any action taken may be only a symbolic response to external legitimacy pressures.

Institutional fields typically face multiple institutional logics—a set of material practices and symbolic constructions that constitute an institutional order’s organizing principles (Friedland and Alford 1991). These institutional logics may complement each other when synergies exist (Montabon et al. 2016), but they can also compete or conflict (Friedland and Alford 1991; Thornton et al. 2012). For example, in the supply network of a sustainability-oriented agricultural cooperative, there can be market-oriented and community-oriented logics, which can co-exist but promote different behaviors (Wu and Pullman 2015). The literature on institutional logics suggests that institutional pressures may lead to heterogeneous rather than homogeneous responses if contending logics co-exist (Thornton and Ocasio 1999). Hence, the presence of conflicting logics has been suggested as one explanation for decoupling (Meyer and Rowan 1977). Yet further examination of how and under what circumstances such trade-offs can become complementary has been called for by Haffar and Searcy (2017).

Decoupling is a well-recognized organization-level response (Boxenbaum and Jonsson 2008). For example, opportunism has been identified as one of the reasons for ill-intentioned non-compliant purchasing behavior (Karjalainen et al. 2008). It can also be due to different development patterns in emerging markets, including a lack of infrastructure and resources, also referred to as ‘institutional voids’ (Khanna et al. 2005). However, more in-depth scrutiny is required into why firms decouple and what allows them to do so (Greenwood et al. 2011). Further, Rogers et al. (2007) highlighted the need to develop an improved understanding of how to cope with decoupling phenomena. More recently, Wilhelm et al. (2016) asserted that there should be a special focus on the implementation and decoupling of socially sustainable practices in suppliers located in developing countries that have a different institutional context. They argued that, because of the inherent nature of social sustainability, it is difficult to separate out the institutional context while studying the implementation of social sustainability.

It has also been suggested that critical events may shift or de-legitimize pre-existing institutional structures (Thornton et al. 2005), creating opportunities to re-evaluate the costs, trade-offs, and benefits of such organizational practices (Sine and David 2003). Institutional logics are important in understanding how attitudes towards the implementation of socially sustainable practices are evolving over time in response to critical industry events. In the context of social sustainability, the two relevant logics are argued to be the economic and social logics. Institutional logics have been studied at the societal, field, and industry level, but very little systematic attention has been paid to how individual organizations react to the multiplicity and incompatibility of logics (Greenwood et al. 2011). Meanwhile, the studies by Rogers et al. (2007) and Wu and Pagell (2011) on competing logics/priorities used static or cross-sectional data. They did not explore how logics/priorities evolve over time. It is argued here to be important to investigate how logics interplay and evolve in developing country supplier firms as events take place to influence the implementation of socially sustainable practices.

Thus, more empirical research into the determinants and deterrents behind implementing ethical practices in developing country suppliers, particularly socially sustainable practices, is necessary (Blome and Paulraj 2013). These major themes from institutional theory lay the foundations for us to explore our research questions concerning how institutional pressures affect the implementation of socially sustainable practices in suppliers situated in challenging institutional contexts; and what explains the decoupling of such formal organizational practices from the day-to-day reality. We contend that this is the logical first step as these antecedents may become contingency factors in future social sustainability-performance relationship studies.

Bangladesh is a major international player in the labor-intensive apparel sector and has been the subject of much public scrutiny. This developing country exported apparel worth $28 billion in 2015/2016, second only to China (McKinsey 2011; BGMEA 2016). However, since 2005, nearly 2000 apparel workers have reportedly died in industrial accidents in Bangladesh (CNN 2013). For example, the Spectrum factory collapse killed 64 people in 2005, the Hameem Group factory fire killed 29 people in 2010, and the Rana Plaza collapse killed 1129 people in 2013. These tragic events have been linked to social failings yet the factories in question had successfully passed various audit checks suggesting a disconnection between the formal adoption of socially sustainable practices and the ground-level reality. Thus, the apparel sector in Bangladesh is suitable for studying the phenomenon of the adoption of ethical practices in a developing country, in particular the adoption (and decoupling) of socially sustainable practices.

We focused on the developing country supplier’s perspective as it was evident from the literature that this required more scrutiny (Zorzini et al. 2015). In particular, we sought to understand how adoption in developing country suppliers is influenced by institutional pressures and why, where relevant, the ground realities have been decoupled from formal, audited practices. Thus, the unit of analysis was the supplier firm interpreted in the institutional context of a developing nation.

Consistent with theoretical sampling, cases were selected opportunistically, i.e., according to their ability to generate new insights that would help in the development of theory (Eisenhardt 1989). Firms were selected from the Bangladeshi apparel industry that were known to supply international customers, i.e., MNCs sourcing from Bangladesh with a minimum revenue of $3 billion. Fourteen supplier firms were approached but were only included if multiple interviewees were available and on-site access was provided. This facilitated triangulation and allowed for observations to inform data collection and interpretation. Further, the set of cases was limited to supplier firms where interviews could also take place in at least one of its international buyers to provide further triangulation. In total, seven suppliers were studied that met the criteria. These seven cases were enough to reach a satisfactory level of theoretical saturation. For example, as data collection approached the end, evidence to support the introduction of new categories of decoupling could not be found. Rather, new evidence only confirmed previously identified categories. Data collection therefore stopped as additional data would not have increased our understanding of the research question (Pagell and Wu 2009).

Data collection took place between December 2011 and May 2015. We followed Yin’s (2009) call for multiple sources of evidence. The primary sources were 19 semi-structured interviews conducted before the Rana Plaza collapse in 2013 with managers from the seven supplier firms (see Table 2). These interviews focused on current social sustainability attitudes and related pressures. For instance, on the issue of institutional pressures faced by supplier firms, we asked what the interviewees thought were the roles of the different institutional actors in exerting pressure and how they would describe this type of pressure. On the decoupling dimension, for example, we tried to determine what kinds of challenges the interviewees were facing in adopting buyers’ social standards and asked them to describe the implementation process. We also probed their motivations and efforts to make their business more socially sustainable. The interviews were supplemented by supplier factory tours and secondary data, e.g., audit reports and news articles. These additional sources were used to inform our understanding of the institutional context and helped us to validate the insights that we received from our primary interviewees.

A further 23 semi-structured interviews were conducted with other relevant institutional actors (see also Table 2). The core set of seven supplier cases was supplemented by evidence from six different buyers from North America and Europe, with apparel sales ranging from $3bn to over $20bn. Evidence was also collected from one local and one international Chamber of Commerce, an apparel trade body, two trade unions, and two NGOs. In addition, two Focus Group Discussions (FGDs: FGD1 and FGD2) were conducted with apparel industry workers with industry experience ranging from 1 to 16 years. FGD1 consisted of 12 workers from 10 different suppliers and FGD2 consisted of 9 workers from 9 different suppliers. These FGDs were conducted in two separate locations, where there was a concentration of apparel suppliers, and they did not include any employees from the seven case firms. These additional sources of data represented many of the institutional actors exerting pressure on implementation. The supplementary interviews provided context and helped obtain both a balanced view and a comprehensive picture of the institutional environment. They also helped us stay objective and triangulate collected information from supplier interviewees with opposing or confirmatory arguments. For example, Table 6 later provides a description of the decoupling coding categories (first column) and the third and fourth columns illustrate how we used the data from these supplementary interviews (buyers and other institutional actors) to triangulate each decoupling factor with the suppliers (second column).

On average, each interview lasted 45 min and most were audio-recorded or accompanied by comprehensive note-taking. Most interviews were conducted in English (except those with the workers and lower management) as English is widely used in business conversations and communication in Bangladesh (Mundi 2011; Uk Trade & Investment 2015). However, during the interviews there would be interjections or further explanations in the Bengali language to add a more informal and conversational style. Where necessary, these were then translated into English by the first author who is fluent in both Bangla and English and thus there were no problems with comprehension. Unclear issues were clarified through follow-up e-mails or telephone calls. Initial interviewees were identified through personal contacts and referrals from a previous interviewee, ensuring participants were both accessible and cooperative (Bryman and Bell 2007). Time was spent building trust with the interviewees to enable ‘frank and open’ discussions, which would otherwise have been difficult given the sensitive nature of the research. This was also aided by the first author and interviewer being a Bangladeshi native.

During the data collection period, the industry suffered a major disaster—the collapse of the Rana Plaza building. This was the world’s deadliest industrial incident since the 1984 Bhopal disaster in India, as well as being the deadliest structural failure in modern human history, killing almost 1200 people. Even though other failures had occurred in the industry, they were nowhere near on this magnitude. Such critical events can trigger institutional change and industry evolution (Thornton et al. 2005) thereby creating prospects for institutional actors to re-evaluate the costs and benefits of existing practices (Sine and David 2003). This offered a unique opportunity to study first-hand the evolution of industry pressures and their effect on the implementation of socially sustainable practices, which would not be possible in environments of stasis. Further, this enabled us to conduct a before-and-after analysis of the change in pressures and the subsequent effect on decoupling. These kinds of opportunities are very rare, especially in qualitative studies.

According to Eisenhardt (1989), if an opportunity presents itself to take advantage of the uniqueness of a specific case and provide new theoretical insights, then researchers can make alterations during the data collection process through “controlled opportunism” (Eisenhardt 1989, p. 539). This is essentially what we did when the Rana Plaza disaster occurred; but it was understandably not possible to obtain perfect symmetry between the before and after data (Huq et al. 2016). New (2015) acknowledged that, in researching unsustainable social practices, researchers might find the data harder to collect and it could be necessary to adopt certain ‘investigative journalist’ approaches. As expected, given the sensitivity of the issues covered, access became more difficult and respondents were more reluctant to share information, and thus, it was not possible to conduct further interviews in all seven suppliers. Through our interactions during the first round of data collection before the Rana Plaza collapse, we were able to gain trust and build comparatively stronger relationships with four of the seven suppliers allowing us to conduct an additional eight post-disaster interviews in these firms. These suppliers were more willing to be involved with the study and, after the Rana Plaza disaster, the relevance of the topic became more evident to them.

This sub-set sample of four cases was however balanced as it covered suppliers showing a low, moderate, and high level of social sustainability performance, as explained fully in Sect. 3.3. It included the best and worst performing suppliers in terms of their level of social sustainability and two moderates (see Table 3). The two polar cases (high vs. low) facilitated building theory on success and failure in implementing socially sustainable practices in challenging institutional contexts (Eisenhardt 1989) while the two moderates helped us cover the entire spectrum and address issues of confirmability and credibility by increasing variance in the data (Wu et al. 2010).

To conduct more substantive and critical work while investigating controversial social failures, it has been suggested that researchers should try to incorporate a wider view, i.e., to include more sources of information than the firms themselves and to try to understand the broader institutional context of the phenomena under exploration (New 2015). Therefore, similar to data collection before the Rana Plaza disaster, the post-Rana supplier interviews were also augmented by ten further interviews in four of the buyers, two NGOs, one labor union, and the Apparel Trade Body (see Table 2). However, we were unable to re-instate the FGDs in this phase and, besides, it would not have been possible to gain collective access to the same group of workers as many were migrant workers from remote regions and the rate of labor turnover is quite high (McKinsey 2011).

Since some of the interviews took place shortly after this critical industry event, there is a chance that interviewees gave ‘knee-jerk’ reactions, leading to biased data through retrospective sense-making by image-conscious informants. However, it is argued that the risk of this is minimized as our data came from a range of highly knowledgeable institutional actors that viewed the focal phenomena from different perspectives (Eisenhardt and Graebner 2007).

Data analysis followed procedures recommended by Eisenhardt (1989) and Miles et al. (2014). Specific techniques were followed to establish rigor through confirmability, dependability, credibility, and transferability. Based on the transcripts, interviewer notes, and secondary data, we first compiled individual case studies for supplier firms, including their level of social compliance and the types of institutional pressure they were facing. As there were no established criteria for measuring the level of social sustainability implementation, we used six proxies identified from the literature (see Sect. 2.2) to gauge the level of social sustainability of supplier firms, including their general attitude towards compliance (e.g., with local laws and buyer/third-party codes of conduct), the depth of their compliance, the social sustainability infrastructure within the firm, and inter-organizational/collaborative approaches to social sustainability (see Table 1). The laggards—Supplier 1 and 2—were rated low on all six criteria, while the leader—Supplier 7—scored high on all six criteria. Those suppliers having a moderate level of implementation—Suppliers 3–6—were rated medium on the majority of criteria (i.e., no more than two criteria rated either high or low). We then expanded the individual case descriptions by adding the factors leading to the suppliers decoupling social sustainability implementation, where relevant. We looked for evidence of barriers, challenges, and impediments that were limiting adoption. This resulted in thick descriptions of each firm’s level of social compliance, institutional pressures, and decoupling factors.

Transcripts from each interview were coded and analyzed using QSR NVivo10^©, which helped to systematically organize the data and be transparent by providing an audit trail (Fielding and Lee 1998; Bryman and Bell 2007). Using such computer-assisted qualitative data analysis software can enhance a researcher’s reflexivity. For example, by using memos, researchers can track the development of their analytical logic and raise their awareness of how they interpreted the data (Woods et al. 2016). After all interviews had been transcribed, the formal coding process began. There was however a degree of overlap between data collection and analysis, which is virtually unavoidable in qualitative research (Pratt 2009). For example, pilot interviews were carried out initially in two of the suppliers. Pilot studies are important as they pre-test research instruments such as the interview schedule; the findings can be used to refine the research questions and generate new themes (Van Teijlingen and Hundley 2001; Bryman and Bell 2007). Interim analysis of the pilot data helped to improve the focus of the research objective, identify possible theoretical lenses, and refine the interview guide.

All of the transcripts were read by the first author and a sample was read by the second author. We looked for patterns based on frequency, commonalities, and differences (Saldaña 2013). A coding frame was developed after discussion, and the transcripts were coded by the first author. Then the second author went through the codes to check for consistency, according to our definitions of the institutional pressures faced by suppliers and the decoupling factors. These were then compared and disagreements discussed. The coding frame was changed when new codes emerged and the above process was followed again until consensus was reached on the final list of codes. Such rigorous coding allowed us to develop analytical categories, which were then conceptualized into theoretical dimensions, as described below.

To begin with, a descriptive, working set of codes was produced based on the themes identified from the literature on institutional pressures. In theory building research, such a provisional specification of codes can facilitate analysis and provide stronger empirical grounding for emergent theory if the codes prove to be important in answering the research questions (Eisenhardt 1989; Saldaña 2013). The sources of coercive, mimetic, and normative pressure on each of the suppliers were identified from the cases. Interviewees within each case were cross-referenced to triangulate the data. The pressures exerted on each supplier were analyzed and cross-case analysis was undertaken, with the suppliers organized according to their perceived degree of social sustainability. The pressures were then further analyzed to arrive at a more interpretive level of coding. For example, formal education was initially identified as a normative pressure leading to adoption (level-one code ‘NOR-EDU’). But as the analysis proceeded, accounts of normative pressure were seen interpretively as involving elements of education, training, and awareness-building of owners/mid-management and workers. Therefore, the initial descriptive code was elaborated to distinguish between the education of workers (‘NOR-EDU-WOR’) and owners/mid-management (‘NOR-EDU-MNG’). The information obtained from interviews was validated by checking documents, focus group findings, and observation notes. Converging perspectives of suppliers and other institutional actors were identified and, where applicable, differences were discussed.

In the second analytical stage, we applied theory building, which enabled us to induce categories of decoupling factors. An initial broad set of codes was first taken from the literature, such as conflict in logics and disparity in socio-economic contexts. Then, analytical categories were inductively derived from the data through an iterative process. At this stage, the interview transcripts were reviewed in detail to detect patterns and regularities. Coding was used to identify the factors leading to decoupling, with data coded as a decoupling factor if it met at least one of the following conditions: (1) it was clearly intended by the supplier as a symbolic action, in that the supplier actively tried to circumvent the buyer’s social standards or made cosmetic changes to pass the social audit in response to institutional pressures; (2) it was perceived as a symbolic action by the researchers; and/or (3) there was disconnect between adopted and actual practices due to local circumstances. Through constant comparison and interpretation, we were able to inductively derive seven subcategories of decoupling in the data (Eisenhardt et al. 2016): (1) conflict in suppliers’ economic and social logics; (2) preparedness to mock comply by suppliers; (3) overlooking of violations by buyers; (4) adversarial relationships with auditors; (5) lack of supply chain visibility; (6) cultural and socio-economic disparity with Western incompatible codes of conduct; and (7) lack of government enforcement of the law. The categories were constructed to reflect as many of the nuances in the data as possible.

The evidence that emerged from the data suggested that these decoupling factors could be described according to three theoretical dimensions: firm-related, supply chain-related, and environment-related factors. For example, we grouped conflict in suppliers’ economic and social logics and preparedness to mock comply into firm-related decoupling factors. Similarly, we grouped cultural and socio-economic disparity with Western codes of conduct and a lack of government enforcement of the law together because these factors were related to the institutional environment in which the suppliers were operating. This process of building theory on decoupling factors is illustrated in Fig. 1. The figure identifies key codes from the literature and links them to the more specific codes inductively derived from the data that feed into the three aggregate theoretical dimensions of decoupling.

A key aspect of qualitative data analysis is removing doubt surrounding the reliability and validity of qualitatively produced findings (Miles and Huberman 1994). Hence, multiple steps have been undertaken to ensure reliability and validity, including data triangulation via site visits, document analyses, and multiple interviews (Stake 1995; Bryman 2008; Yin 2009). Finally, cross-case analysis was undertaken to search for patterns, i.e., similarities and differences between the coded categories across the seven suppliers. For example, some categories such as buyer coercive pressure revealed no clear patterns, but others such as normative pressure through the education of workers led to important patterns of within-group similarity and cross-group differences. This systematic analysis of our case data aided in ensuring rigor thereby increasing confidence in our findings.

As we have seen in the previous section, the most significant pressure on suppliers’ social sustainability performance came from buyers, who introduced stringent codes of conduct. Many of the suppliers interviewed had successfully become approved suppliers to major international retailers, allowing them to continue to win orders. Nonetheless, problems clearly remained, as was evident from the Tazreen factory fire just 5 months before the Rana Plaza collapse.

In this section, we unbundle the factors leading to decoupling into firm-, supply chain-, and environment-related factors. Table 5 shows the types of decoupling factor evident in each supplier while Table 6 provides a description of the decoupling coding categories and illustrates through quotes from suppliers, buyers, and other institutional actors how each decoupling factor was triangulated.

After the Rana Plaza collapse, a new phenomenon emerged—major collective action led by groups of buyers. Two groups in particular were formed. Accord consisted of over 160 mainly European buyers that now worked together with global and local trade unions, supported by NGOs. The other group, Alliance, brought together 26 North American apparel brands and trade associations. The two groups use similar standards to conduct fire, electrical, and structural safety audits of suppliers, while Accord has also pledged to contribute towards the costs of improvements. Three out of the four suppliers interviewed after the collapse fell under the governance of at least one of these groups. The groups started to make their audit reports publicly available, placing greater coercive pressure on suppliers—essentially naming and shaming non-compliant factories.

Collective action meant the repercussions of failing an audit were severe: not just one, but a whole group of buyers would no longer source from a supplier. Factories that did not meet the standards were effectively being shut down. Supplier 3’s DMD conceded: “adhering to the new standards is a matter of survival. If you do not pass, either you lose the business or the worst case scenario is you are forced to shut down.” Similarly, Supplier 6’s Director explained: “We have no option but to make costly changes. It might cost us around $350,000… but it is a requirement I have to fulfil.” Such collective, coercive pressure by buyers served to make social and economic logics in developing country suppliers more closely connected. Thus, social logic was heightened, even if the economic logic continued to dominate. These findings extend previous research on how powerful buyers can promulgate change in suppliers (Benton and Maloni 2005; Yu 2008; Marshall et al. 2016) and lead to our first proposition:

Collective coercive buyer pressure also helped to mitigate decoupling caused by the lack of law enforcement that previously undermined implementation. Supplier 3’s DMD explained: “The biggest change was brought about by Alliance and Accord. If the buyers, suppliers, or the government for that matter, wanted to do this individually through their own inspections they would never have been able to do what these two organizations have done.” In effect, the two groups of buyers were playing a quasi-governmental role since the government was unable or unwilling to fulfil this function. This leads to the following proposition:

Suppliers with a low level of social sustainability implementation (e.g., Supplier 1) appeared to be unable to overcome the apparent conflict between the social and economic logics, even after the disaster. Further, they did not possess the capabilities to make the more expensive improvements now demanded collectively by buyers, who were also trying to reduce their supplier bases and levels of risk exposure by procuring from (fewer) larger and more compliant suppliers only. The number of suppliers reduced from 5876 in 2012/2013 to 4328 in 2015/2016 (BGMEA 2016). In fact, the Director of Supplier 1 disclosed to us that they had decided to end their business as a direct result of their inability to cope with increased buyer requirements after Rana Plaza. She stated: “Rana Plaza had a direct impact on our decision of shutting down the business. We do not have our own building [it is rented] and some of the compliance issues like having extra staircases cannot be solved. So we will fail every audit. They require 50,000 gallons of water reserve, we don’t have this. Unless we have our own building and are willing to make this kind of investment [for social compliance], our main buyer will not continue with us…After Rana they [buyers] want us to be more socially compliant, but they are not willing to pay us for it…The trend will be that only the big players who can bear the cost of increased compliance requirements will survive and the smaller suppliers will shut down” This example shows that Supplier 1 was not able to survive in the industry because of its unwillingness to meet the greater social sustainability expectations of buyers, and this unwillingness was largely due to perceived conflict with its more dominant economic logic. This leads to the following proposition:

Our analysis of the case data before the collapse of Rana Plaza showed that one of the reasons for firm-related decoupling is mock compliance. This critical finding is connected to the larger literature stream on the effectiveness (or lack thereof) of codes of conduct and, in particular, the audits that many customers require (Huq et al. 2014; Egels-Zandén and Lindholm 2015). However, after the Rana Plaza disaster, suppliers were incentivized to reach minimum acceptable compliance standards. One of the reasons was amplified mimetic pressure through increased competition for orders as buyers backed away from procuring from less compliant suppliers, and some buyers exited the Bangladesh market altogether. Thus, orders became harder to secure, and this made the risks of mock compliance too great for the reward. This leads to the following proposition:

After the Rana Plaza disaster, buyers (including via coalitions/horizontal collaboration) invested in the education and training of their own staff in Bangladesh and in that of their suppliers at all levels—owners, middle managers, and workers. For example, buyers used external consultants to train and develop supplier fire and building safety capabilities. Buyer 2’s Compliance Manager explained: “my firm flew me to the UK to attend training programmes on fire safety. We are now providing our suppliers with expert help.” Meanwhile, Supplier 3’s DMD described how buyer groups were educating and training his workers in stages. They first came with tablet computers and gave workers multiple choice questions to test their level of fire safety knowledge then shared the results with the management. He explained: “They asked the workers what they will do in case of a fire. Most answered that they would hide under the table! ... We were shocked as our HR department was ‘training’ them. We found that our training was not at all effective and it was not communicated well.” As a result, Supplier 3 sent around 20% of their workforce for fire safety training before becoming certified fire fighters. Through education and training, these workers became more aware of social compliance issues and were able to exert normative pressure on the suppliers. Buyer groups also tested and trained mid-level managers, nurses, and electricians. But, according to the DMD of Supplier 3, the most important change was the education of the owners, which infused in them the need to have higher social standards. These measures decreased conflict between the social and economic logics in suppliers. This discussion leads to our next proposition.

Compliant factories, particularly proactive, early adopters that had built their competitive strategy around social sustainability now began to be rewarded. Indeed, once a supplier had invested in being socially responsible, it began to reap rewards through worker productivity and customer orders, which led to an enhanced reputation, attracting new customers, and so on. In particular, Supplier 7 had the most heightened social logic in our sample, winning numerous awards and paying its workers 20% above the industry average. Supplier 7 understood that the economic and social ideologies could complement each other in the long term and found improving standards actually increased productivity via worker retention and reduced absenteeism. Being socially sustainable was improving the company’s reputation and helping to attract more and better customers. In fact, the collapse of the Rana Plaza led to an influx of orders from like-minded buyers who not only offered better prices to compliant factories but also involved their suppliers in community projects, collaborating with various NGOs and donor agencies. This kind of stakeholder collaboration created extra normative pressure on such proactive suppliers. This discussion leads to our final proposition: