The Challenge of Implementing Voluntary Sustainability Standards: A Dynamic Framework on the Tension between Adherence and Adaptation

Voluntary sustainability standards (VSS) aim to encourage ethical behaviors of organizations, yet studies show that many VSS adopters do not live up to these promises. Existing literature typically attributes the reason for this ineffectiveness to either policy–practice decoupling, owing to a lack of adhering to VSS requirements, or means–ends decoupling, owing to a lack of adapting to the local context. However, little is known about how the contradictory needs of adherence and adaptation evolve throughout VSS implementation. Building on the knowledge transfer literature, we develop a dynamic conceptual framework that distinguishes two phases of VSS implementation. Specifically, we theorize how tensions emerge in the transition between phases since the first phase primarily calls for adherence, whereas the second calls for adaptation. Applying this framework, we develop propositions to illustrate how these tensions relate to different VSS characteristics: stringency, enforcement, and scope. The article concludes with implications and future research directions for VSS scholarship.

assess, measure, audit, and/or communicate the social and environmental behavior and/or performance of firms" (Gilbert et al., 2011, p. 24).
Yet evidence suggests that VSS do not live up to the promises and that reaching the desired sustainability goals is the exception, not the norm (Dietz, Chong, Grabs, & Kilian, 2020;Giuliani, Ciravegna, Vezzulli, & Kilian, 2017;Waldman & Kerr, 2014). A prominent explanation proposed for their ineffectiveness is a lack of adherence to the VSS requirements. Policy-practice decoupling happens when an organization formally adopts the VSS but chooses not to properly implement the requirements, taking advantage of weaknesses in the monitoring process (Behnam & MacLean, 2011;Brunsson, Rasche, & Seidl, 2012;Christmann & Taylor, 2006). Yet too strict adherence to the requirements can backfire, resulting in means-ends decoupling, which happens when the organization complies with the VSS requirements but still fails to obtain the desired outcomes, because rigid requirements cannot encompass the adaptations needed to reach the goals given context specificities (Bromley & Powell, 2012;Wijen, 2014). Therefore a tension emerges between the need to foster adherence to VSS requirements to reduce the risk of policypractice decoupling and the need to allow for adaptation to reduce the risk of means-ends decoupling (Christensen, Morsing, & Thyssen, 2017;Rasche, 2010;Sandholtz, 2012;Wijen, 2014).
Although scholars are cognizant of this tension (Brunsson et al., 2012;Christensen et al., 2017;Gilbert et al., 2011;Wijen, 2014) and have called for further research into the dynamics of standards (Brunsson et al., 2012), existing literature of VSS implementation has not examined whether and how the tension between adherence and adaptation evolves over time. Most decoupling studies, owing to their research design, portray VSS implementation as an act and do not include a temporal dimension (Aravind & Christmann, 2011;Barrientos & Smith, 2007;Giuliani et al., 2017). Such a static perspective clashes with the evidence from a few longitudinal studies on VSS implementation that reveal how organizational members implement and integrate VSS requirements gradually over time, while dealing with challenges that emerge during the process (Boiral, 2007;Egels-Zandén, 2014;Lazaric & Denis, 2005;Sandholtz, 2012). However, even longitudinal studies do not distinguish different phases of VSS implementation (for an exception, see Tampe, 2021) or integrate the tension between adherence and adaptation into their discussion. Overlooking the dynamics of this tension is problematic, as it also disguises potential ways to manage it more effectively.
Addressing this oversight, we ask the following research question: How does the tension between adherence and adaptation evolve dynamically throughout the implementation of VSS? To answer this question, we develop a conceptual framework to theorize the dynamics of VSS implementation, integrating insights from the knowledge transfer literature (D'Adderio, 2014;Szulanski, 2000Szulanski, , 1996Szulanski, Winter, Cappetta, & Van den Bulte, 2002). This literature is particularly suitable to our objective of theorizing VSS implementation for two reasons. First, it has theorized knowledge transfer as a process, and more specifically, it discusses how the needs for adherence and adaptation develop over time as a key insight on which we build. Second, this literature is relevant to VSS because VSS implementation 297 Implementing Voluntary Sustainability Standards requires that knowledge circulate from the source of knowledge (i.e., the standardsetter) to the recipient of knowledge (i.e., the adopter). The knowledge transferred, in the VSS context, refers to both the VSS requirements (for instance, norms, rules, guidelines, formal prescriptions, or best practices) and the knowledge to implement them (for instance, knowing how to make changes in materials, production, and organizational processes).
Building on the knowledge transfer literature, we delineate two phases of VSS implementation-adoption and integration-and conceptualize how the needs for adherence and adaptation evolve along these phases. We propose that, whereas the first phase calls primarily for adherence, the second phase requires a greater degree of adaptation, albeit with counterbalancing mechanisms to equilibrate the competing needs within each phase. Our framework reveals that tensions emerge particularly when the emphasis shifts from adherence in the adoption phase to adaptation in the integration phase. Specifically, we identify three tensions-of proximity, autonomy, and interpretability. Finally, we develop propositions to illustrate how our framework applies to different types of VSS based on three characteristics, namely, stringency, enforcement, and scope.
Our article contributes to the multidisciplinary scholarly conversation on VSS in three ways. First, by introducing insights from the knowledge transfer literature, we move beyond the dominant static perspective on VSS and instead theorize VSS implementation as a process evolving dynamically. Second, our framework provides greater clarity on the tensions that thwart VSS implementation and explains how certain VSS characteristics and their combinations enhance VSS effectiveness. Finally, we contribute to the literature on decoupling (Aravind & Christmann, 2011;Brunsson et al., 2012;Wijen, 2014) by illustrating how policy-practice and meansends decoupling are interrelated and occur over time. These contributions provide relevant pathways for scholars, practitioners, and standard-setters alike who look for solutions to the social and environmental challenges present in global business.
In the next section, we introduce our theoretical foundation from the literature on VSS implementation and decoupling as well as knowledge transfer. In the subsequent section, we develop our dynamic framework, proposing two phases of VSS implementation with their respective needs for adherence and adaptation and the tensions emerging between the two phases. We then move to illustrate how the conceptual framework applies to different characteristics of VSS, and finally, we discuss our contributions as well as boundary conditions and directions for future research.

VSS Implementation and Decoupling
VSS are a form of governance developed to hold organizations accountable for their practices, as they represent a way to evaluate the environmental, ethical, or social performances of an organization and to communicate them to third parties (Gilbert & Rasche, 2007). VSS are developed by international organizations, multistakeholder initiatives, nongovernmental organizations (NGOs), industry associations, or 298 Business Ethics Quarterly companies and can either emit certifications upon verifying the implementation of requirements or rely on voluntary disclosure and self-reporting (de Bakker, Rasche, & Ponte, 2019;Gilbert et al., 2011;Pope & Lim, 2020). Examples of schemes that emit certifications are ISO 14001, an environmental management system (Aravind & Christmann, 2011;King, Lenox, & Terlaak, 2005); SA8000 for labor rights (Gilbert & Rasche, 2007); and Fairtrade for a fairer distribution of value within supply chains (Schuler & Christmann, 2011). Examples of voluntary disclosure and self-reporting schemes are the United Nations Global Compact (UNGC), which stipulates ten universally accepted principles including both social and environmental dimensions (Leisinger, 2007;Rasche & Kell, 2010), and the Global Reporting Initiative (GRI) for disclosing nonfinancial performance (Lim & Tsutsui, 2012;Shanahan & Khagram, 2006). However, VSS have often failed to result in more responsible social and environmental behaviors. For instance, in the coffee value chain, neither Fairtrade nor organic standards have succeeded in preserving biodiversity in the highlands of Chiapas, Mexico (Philpott, Bichier, Rice, & Greenberg, 2007). The effects of VSS are disappointing for social and labor practices too (Giuliani et al., 2017). In Kenya, coffee producers' marginal income increased only 10 percent due to participation in VSS programs, and in Uganda, only Fairtrade, out of the major VSS programs, has been found to improve living conditions (Van Rijsbergen, Elbers, Ruben, & Njuguna, 2016). Similar disappointing results have been documented in various industries, including soybean production (Waldman & Kerr, 2014), fisheries (Tolentino-Zondervan, Berentsen, Bush, Idemne, Babaran, & Lansink, 2016), and apparel (Locke, Amengual, & Mangla, 2009).
The traditional explanation for such limited effectiveness is that organizations decouple the formal adoption of VSS from the implementation of the requirements (Brunsson et al., 2012;Christmann & Taylor, 2006;King et al., 2005). Policypractice decoupling happens when organizations adopt VSS to grasp the benefits in terms of legitimacy and signaling benefits, without bearing the costs of (full) implementation (Aravind & Christmann, 2011;Egels-Zandén, 2014;King et al., 2005). For instance, Aravind and Christmann (2011) conclude that a low-quality implementation of ISO 14001 requirements explains the lack of environmental performance improvements. Similarly, in his longitudinal study of Chinese toy suppliers, Egels-Zandén (2014) found that external pressure for adherence to requirements and a less ceremonial auditing process are essential to foster the implementation of VSS requirements and to reduce the number of violations. Enforcement mechanisms, such as monitoring and sanctions, are recommended to overcome this type of decoupling, assuming that this would motivate adopters to adhere fully to the requirements (Aravind & Christmann, 2011;Behnam & MacLean, 2011;Christmann & Taylor, 2006;Egels-Zandén, 2014).
Recently, some authors have pointed out that the focus on adherence is not necessarily desirable and may undermine rather than enhance the effectiveness of VSS (Christensen et al., 2017;Rasche, 2010Rasche, , 2012Wijen, 2014). A second type of decoupling, called means-ends decoupling, spurs from too strict an adherence to the letter of VSS requirements, which may not be adequate for leading to the desired 299 Implementing Voluntary Sustainability Standards social and environmental outcomes in the context of implementation (de Bakker et al., 2019;Christensen et al., 2017;Wijen, 2014). While VSS aim at defining universal rules to deal with global challenges, implementation happens in idiosyncratic local contexts that do not easily align with global rules (Heimer, 2013;Huising & Silbey, 2011) and that differ significantly in their technologies, ecological and social systems, resources, and capabilities (Corredoira & McDermott, 2014;Perez-Aleman, 2013, 2011. Therefore, even when full compliance is achieved, implementing the requirements might not yield the desired goals, given the context specificities (Wijen, 2014). For example, the amount of water needed for irrigation depends on the soil and climate conditions of the specific region (Wijen, 2014), requiring an adaptation of outcome-based regulations of water usage to different contexts. To overcome means-ends decoupling, scholars propose carefully adapting requirements to fit the specificities of a given implementation context (Brunsson et al., 2012;Christensen et al., 2017;Wijen, 2014).
Read together, there is a tension between the pressure for adherence and the pressure for adaptation. On one hand, VSS have been developed to diffuse standardized practices, and adherence is key to achieving this objective. On the other hand, reaching uniformity across time and space is often impossible, as each context represents unique characteristics (Thévenot, 2009). The literature on VSS has started acknowledging this tension and the need to balance adherence and adaptation to foster the effectiveness of VSS (Brunsson et al., 2012;de Bakker et al., 2019;Rasche, 2010Rasche, , 2012. Solutions to address this tension include recommendations for dialogue and participation to balance divergent needs by involving different stakeholders in the decision process in a climate of fairness, consensual orientation, and transparency (de Bakker et al., 2019;Mena & Palazzo, 2012;Overdevest & Zeitlin, 2014).
Although VSS and decoupling studies have highlighted important aspects of VSS implementation, they often adopt a static perspective, leading to calls for more attention to the dynamics of standards (Brunsson et al., 2012). The pressure for adherence and adaptation, and the tension between these two demands, is implicitly represented as constant from the time the VSS is formally adopted until it is discontinued. This static perspective is evident also in the methodology used by most empirical studies, which are often cross-sectional and capture VSS implementation at a specific point in time (Aravind & Christmann, 2011;Barrientos & Smith, 2007;Giuliani et al., 2017), which is recognized as a major shortcoming in VSS theorization (Egels-Zandén, 2014).
In contrast with a static view of VSS implementation, a few longitudinal studies reveal how VSS implementation is not an act but a long and complex process (Boiral, 2007;Egels-Zandén, 2014;Lazaric & Denis, 2005;Sandholtz, 2012;Tampe, 2021). Scholars have shown that the adopters of VSS often face steep learning curves for disrupting their old practices and adapting them to the ones required by VSS (Huising & Silbey, 2011;Perez-Aleman, 2011). For instance, a longitudinal study on the implementation of the UTZ standard in Brazil illustrates how cocoa producers needed first to build a new health and safety practice and later to keep supervising the new practice until it became habitual (Tampe, 2021).

Business Ethics Quarterly
Although these studies do not theorize the evolution of adherence and adaptation needs, they provide evidence that the challenges of implementing VSS evolve over time. Therefore a dynamic view is needed to investigate how decoupling risks, and the related need for adherence and adaptation, affect different phases of VSS implementation. We argue that a closer look at the literature on knowledge transfer (D'Adderio, 2014;Szulanski, 1996Szulanski, , 2000 gives valuable insights for VSS implementation, as this literature illustrates how the tension between adherence and adaptation manifests dynamically over the knowledge transfer process. Moreover, VSS also function as a tool to transfer the knowledge about requirements and how to implement them from standard-setters to adopters, either directly or indirectly, via auditors or other third parties that monitor the VSS implementation (Giuliani et al., 2017;Perez-Aleman, 2011Tolentino-Zondervan et al., 2016).

Knowledge Transfer as a Lens for VSS Implementation
The literature on knowledge transfer provides the conceptual tools to unpack the evolution of the tension between adherence and adaptation over time (D'Adderio, 2014;Szulanski & Jensen, 2006), referred to in this literature as the "replication dilemma" (Winter & Szulanski, 2001: 737). On one hand, the knowledge transfer should adhere faithfully to the original template because altering a proven knowledge template without having an accurate understanding of the cause-effect relationships involved is likely to undermine the effectiveness of the new knowledge (Jensen & Szulanski, 2004;Winter, Szulanski, Ringov, & Jensen, 2012). On the other hand, the pressure to adhere strictly to the template can undermine the knowledge transfer effectiveness because it prevents local innovation and adaptation to the local context, which are crucial, especially for cross-border transfers, where the characteristics of the recipient environment are different from those of the source environment on multiple dimensions, such as culture, regulations, and market forces (Ansari, Fiss, & Zajac, 2010;Onkvisit & Shaw, 1987).
Unlike the VSS literature, the knowledge transfer literature adopts a dynamic perspective on this tension by distinguishing two phases of knowledge transfer implementation (Chandler, 2014;D'Adderio, 2014;Kostova, 1999): 1) an initial phase (often referred to as the transfer phase), which spans from when the source of knowledge initiates the knowledge transfer to when the recipient starts to implement the new knowledge, and 2) a second phase (often referred to as posttransfer or the integration phase), when the knowledge recipient continuously uses the newly acquired knowledge and integrates it with existing routines (D'Adderio, 2014;Szulanski, 1996Szulanski, , 2000.
In the initial phase, the success of the transfer depends on the capacity and willingness of the actors involved to bridge the knowledge gap (Szulanski, 1996(Szulanski, , 2000Szulanski et al., 2002). As Gondo and Amis (2013) point out, the main challenge in this phase is the recipient's lack of acceptance of the new knowledge, leading to a conscious decision to decouple practices from the knowledge transferred. Problems arise when the source of knowledge lacks the motivation or authority to transfer the knowledge, when the recipient lacks absorptive capacity (i.e., the capacity to acquire and retain new knowledge) or perceives the source of knowledge as unreliable, or when the knowledge transferred is unproven or ambiguous (Szulanski et al., 2002). Overcoming such barriers requires an emphasis on adherence to the knowledge template (D'Adderio, 2014) and on monitoring the faithful implementation by the source of knowledge (Szulanski, 1996(Szulanski, , 2000. In the second phase, the recipient is responsible for the continuous use of the new knowledge until it is fully integrated into the organization's routines (Kostova, 1999). Empirical evidence from knowledge transfers within eight multinational firms suggests that this task is far from trivial (Szulanski et al., 2002). Inconsistencies between the new knowledge and existing organizational practices inevitably emerge, as the knowledge transferred is developed in one context and implemented in a different one (Szulanski, 1996). The recipients' role becomes more prominent, as the recipients need to adapt the new knowledge to fit with the context. As Gondo and Amis (2013) suggest, the lack of fit between the new knowledge and existing practices emerging in this phase leads to a different type of decoupling, where the recipients try to implement the new practices but are not able to gain the expected benefits from the new knowledge.
Importantly, recent findings on the microdynamics of knowledge transfer show that the recipients of knowledge enact the contrasting goals of adherence and adaptation simultaneously, by activating in each phase the more prominent goal and backgrounding the other, which is not, however, entirely suppressed (D'Adderio, 2014). In both phases, both adherence and adaptation are pursued, albeit to a different extent. In the initial phase, counterbalancing mechanisms for adaptation complement the focus on adherence, whereas in the later phase, the mechanisms supporting adaptation gain more ground, while counterbalancing adherence mechanisms help to control alignment of adaptations of the new knowledge with the intended knowledge transfer (D'Adderio, 2014).
In sum, the knowledge transfer literature provides important conceptual underpinnings for VSS implementation by theorizing the phases and their different pressures for adherence and adaptation. However, further conceptual work is needed to account for the differences between the contexts of investigation of the knowledge transfer literature, mainly, the transfer between the headquarters and subsidiaries of multinational companies, and the VSS context. The latter is generally characterized by high field opacity and causal complexity (Wijen, 2014), which results in an even more prominent tension between adherence and adaptation. Therefore we know little about how this tension might play out dynamically in the VSS context. By homing in on key elements of the transfer-the source of knowledge, the recipient of knowledge, and the knowledge itself-we aspire to elucidate how this tension manifests in the context of VSS implementation.

A DYNAMIC FRAMEWORK OF VSS IMPLEMENTATION
In this section, we develop a framework to illustrate how the tension between adherence and adaptation dynamically develops during the implementation of VSS. First, we build on the knowledge transfer literature and distinguish two phases 302 Business Ethics Quarterly in the implementation of VSS, namely, adoption and integration-the first characterized by a higher risk of policy-practice decoupling, thus calling for higher adherence, and the second by a higher risk of means-ends decoupling, thus calling for higher adaptation. Crucially, these two phases are not to be understood as clearly separable and distinguishable phases but rather as intermingled, with a transition between the two. We then discuss what mechanisms support the primary pressure for adherence and adaptation in the two phases, respectively, and the counterbalancing mechanisms required, before elaborating on the tensions that emerge in the transition between phases. Figure 1 shows our framework, illustrating how tensions emerge as the actors implementing VSS try to overcome both policy-practice and means-ends decoupling.

The Adoption and Integration Phases of VSS Implementation
The first phase of VSS implementation, which we call adoption, starts with the recipient's decision to adopt the VSS and ends with the formal adoption. If policypractice decoupling does not occur, this phase is marked with the initial implementation of the requirements and thus with compliance. For certification schemes, the end of this phase coincides with the certification award certifying that all requirements have been implemented, while for VSS without a formal adoption marker in time, the transition to the second phase is likely more gradual. In this case, we consider the adoption phase to end when the adopter communicates the formal adoption of the VSS and the implementation of the requirements. During the adoption phase, most of the knowledge embedded in the VSS requirements is transferred from the source of knowledge to the recipient organization. For example, in the case of ISO 14001, the adopting organization receives a document from ISO with an explanation and rationale for each requirement. Subsequently, the organization must assess the current status of the organization, develop an implementation and training plan to meet the requirements set by the VSS, and then pass a site and document review by an accredited monitoring and assessment body that decides whether to emit the ISO certificate (Font, 2002). Concomitantly, buyers, development agencies, and NGOs can act as complementary sources of knowledge in VSS implementation, particularly for labor-and production-related VSS requirements (Locke et al., 2009;Tampe, 2018).
In line with the knowledge transfer literature, we argue that, in most cases, the adoption phase implies a high risk of policy-practice decoupling and therefore a greater pressure for adherence. The adopters might be tempted to formally adopt the VSS for reputation and signaling benefits without meeting all requirements, especially when the implementation of the VSS implies substantive upfront investments without a clear payoff (Goedhuys & Sleuwaegen, 2013;Kumar, Thapa, Roy, & Joshi, 2017), when there is great distance between the requirements and adopters' existing practices (Perez-Aleman, 2011), or when the adopters are isolated (Tampe, 2021). In such cases, knowledge, resources, and time are needed to enhance existing technology, infrastructures, and reporting procedures; to increase salaries; or to improve working conditions and training for the employees (Behnam & MacLean, 2011;Yeung & Mok, 2005). Moreover, as the requirements are initially implemented, 303 Implementing Voluntary Sustainability Standards

304
Business Ethics Quarterly there is likely resistance from some organizational members to accept and modify their existing practices due to the discomfort generated by the required changes in practices (Lazaric & Denis, 2005). All these factors create incentives for the organization to formally adopt the VSS without properly implementing the requirements, calling for mechanisms to foster adherence. In comparison, the means-ends decoupling and the pressure for adaptation are often less prominent in the adoption phase, where the focus is on adhering to the requirements to formalize the adoption. The first reason for this is because the effectiveness of the newly implemented practices and the related risk of meansends decoupling will take time to manifest. To improve the effectiveness of VSS in the local context, the recipient needs gradually to become familiar with the requirements and to understand the underlying cause-effect relationships and unintended consequences of the VSS implementation (Sutter, Kistruck, & Morris, 2014). Second, leaving ample room for adaptation when there is a high risk of policy-practice decoupling is problematic, as organizational members may adapt certain requirements in unfaithful ways owing to their resistance to change or unwillingness to invest in compliance (Boiral, 2003). However, the risk of means-ends decoupling is not absent. When taken to an extreme, adherence becomes counterproductive, for example, when a global source of knowledge is too far removed to grasp the challenges from recipients' reality (Heimer, 2013), when requirements visibly misfit the local context (Wijen, 2014), or when requirements prescribe every detail of implementation and risk stifling all local innovation and demotivating adopters (D'Adderio, 2014; Ponte & Ewert, 2009). All these instances reduce the effectiveness of VSS in the local context. Therefore, though the pressure for adherence is predominant, especially when complying with the VSS requires substantial changes, this phase also benefits from counterbalancing mechanisms for adaptation.
The second phase, which we call integration, begins after the implementation of the requirements and continues until the requirements are fully integrated by the adopter as new and lasting routines of the organization. While continuously enacting the required practices, the members of the adopting organizations are likely to make adaptations in the way requirements are implemented over time to improve the efficacy and effectiveness of implementation. The relevant knowledge in this phase is less about compliance with requirements and more about how to integrate them effectively into the local context to build long-lasting routines. A study of how a French meat producer, Defial, implemented the ISO 9002 standard illustrates how, in the four years following the certification award, the organization underwent a series of organizational adaptations, even if the company was formally complying with the requirements of the standard during the first months (Lazaric & Denis, 2005). For example, Defial adjusted its quality control process by internalizing the analysis lab and hiring more employees and thus was able to improve and routinize the ISO-required practices over time.
In line with the knowledge transfer literature, we argue that, in this phase, the risk of means-ends decoupling and the related pressure for adaptation become prominent, as the effectiveness of the new practices in the context is evaluated and potential unintended consequences of implementing the requirements are likely to 305 Implementing Voluntary Sustainability Standards emerge, especially when the standard-setter is not familiar with the local implementation context. For instance, Fairtrade prescribes cooperatives as a way to empower producers and to instill workplace democracy, but investigative journalists have criticized this requirement as aggravating problems with exploitation and abuse in the African context (Wijen, 2014). By the time of the integration phase, adopters are more likely to have assessed the efficacy of VSS implementation and to have gained the competence and experience needed to adapt the requirements accordingly in the local context. Therefore a strict focus on adherence in the integration phase may not only be unnecessary but can even undermine the effectiveness of VSS.
The risk of policy-practice decoupling, although still present, is less prominent than in the previous phase, especially for the types of requirements that have higher initial implementation costs compared to the costs of continuous implementation. For instance, updating the technologies and systems used, providing employee training, and communication efforts to overcome initial resistance are upfront costs that the adopter must often bear when the requirements are initially implemented. Substantive upfront investment reduces the incentives to decouple VSS adoption and implementation after the initial implementation and lessens the pressure for adherence mechanisms. However, there is always a risk that newly implemented practices may be discontinued as initial attention to VSS requirements and supervision fatigue wear off (Staats, Dai, Hofmann, & Milkman, 2017), because more space for adaptation of requirements may become a way to hide a substantive dismissal of the new practices before their integration (Bowler et al., 2017;Lazaric & Denis, 2005). Therefore, although the pressure for adaptation is predominant, this phase also benefits from counterbalancing mechanisms for adherence.
In sum, we underlined that the risk of policy-practice decoupling is particularly relevant in the adoption phase, calling for adherence mechanisms, and less prominent, albeit not absent, in the integration phase. In contrast, the risk of means-ends decoupling becomes more relevant in the integration phase, calling for adaptation mechanisms, whereas it is less salient during the adoption phase. This dynamic applies particularly when two boundary conditions, common to many instances of VSS implementation, are met: first, when substantive upfront investment is required for the recipients and the implementation of VSS requirements implies a significant change in organizational practices (i.e., the VSS adoption is not merely a recognition of a set of practices already implemented), and second, when the source of knowledge has an inaccurate picture of the implementation context and lacks the capacity to guide the local adaptation and integration of practices, as is often the case for VSS that are implemented in widely differing contexts. Next, we discuss what mechanisms support the focus on adherence and adaptation in the two phases before turning to the specific tensions that occur between these phases.

Mechanisms in the Adoption Phase
Drawing on the knowledge transfer literature, we distinguish a set of mechanisms that foster adherence in the adoption phase of VSS implementation, referring to the source of knowledge (i.e. the standard-setter, either an international organization such as the ISO, a multistakeholder initiative, or a large buyer, as is the case for 306 Business Ethics Quarterly company-led VSS), the recipient of knowledge (i.e. the adopter), and the knowledge transferred (i.e. the VSS requirements and the knowledge on how to implement them), respectively. First, the literature on knowledge transfer underlines how the authority and motivation of the source of knowledge are important to foster the acceptance of the new knowledge in the initial phase (Szulanski, 1996(Szulanski, , 2000. In the VSS context, we argue that a global source of knowledge provides pressure for adherence. With a global source of knowledge, we mean standard-setters that have global exposure and visibility, such as the United Nations. Accordingly, the VSS under the aegis of highly visible standard-setters, too, are particularly subject to public and civil society inspection, evaluation, and criticism in case of ineffectiveness (Haack & Rasche, 2021). Such public exposure creates reputational risk if ineffective or poor implementation is discovered (Donaghey, Reinecke, Niforou, & Lawson, 2014;Josserand & Kaine, 2016;Schuler & Christmann, 2011) and consequently increases pressure to foster adopters' compliance through stricter adherence mechanisms (Delmas & Montiel, 2009 (Weitzman, 2006), the Fairtrade Labelling Organisations International (FLO) undertook radical changes in its governance and requirements and strengthened its enforcement mechanisms to restore and maintain its credibility and reputation (Bennett, 2013(Bennett, , 2016. Similarly, other global standardsetters have strengthened their adherence mechanisms to increase their legitimacy, as the UNGC did with introducing the "communication of progress" policy to monitor implementation (Haack & Rasche, 2021).
Second, the knowledge transfer literature posits that, in the initial phase, the recipient of knowledge needs to possess absorptive capacity and perceive the source as trustworthy to accept the new knowledge (Szulanski, 1996(Szulanski, , 2000. In the VSS context, trust and absorptive capacity are enhanced when the recipient of knowledge interacts frequently with the source of knowledge, mainly the standard-setter. A recipient of knowledge close to the source of knowledge increases the recipient's pressure to adhere to the requirements. With a close relationship, we mean that the recipient and the source have frequent interactions, for instance, when the standardsetter is directly involved in monitoring and supporting the implementation of requirements. Close relationships with the knowledge source reduce the risk of policy-practice decoupling for two reasons. First, they foster better communication and problem solving, higher acceptance of advice from the source of knowledge, and lower rejection of not-well-understood requirements (Josserand & Kaine, 2016;Locke, 2013;Pipkin & Fuentes, 2017). When such relationships are in place, the source of knowledge allocates more effort to transferring the relevant knowledge and explaining the requirements and the steps needed for their implementation, stimulating the absorptive capacity of the recipient. Second, frequent interaction with the source of knowledge increases the risks and cost of noncompliance for the recipient of knowledge (Delmas & Montiel, 2009;Riisgaard & Hammer, 2011;Tolentino-Zondervan et al., 2016). For instance, a study of company-led VSS in 307 Implementing Voluntary Sustainability Standards Chinese toy suppliers found that policy-practice decoupling is reduced by frequent interactions between the adopters and the standard-setter (Egels-Zandén, 2014).
Finally, the knowledge transfer literature underlines that the recipient's understanding of the transferred knowledge is a necessary condition for its acceptance (Szulanski, 1996(Szulanski, , 2000. Similarly, in the context of VSS, when the knowledge transferred is clear and unambiguous (i.e., when the requirements and the steps to implement them are clear and detailed), the risk of policy-practice decoupling is lower (Behnam & MacLean, 2011;Egels-Zandén, 2014;Giuliani et al., 2017). The reasons are twofold. First, clear VSS requirements increase adherence by providing step-by-step guidelines that are easy to understand and enforce without requiring a high level of preexisting knowledge. For instance, a study on labor rights found that VSS requirements on health and safety, such as protective equipment, fire safety, or drinking water, were often implemented, whereas VSS requirements on freedom of association or collective bargaining had little or no impact on practice, partly because the latter do not provide clear and detailed enough guidelines for processes or outcome expectations (Barrientos & Smith, 2007). Second, setting unambiguous expectations related to compliance also reduces the possibility of opportunistic interpretations where certified organizations can take advantage of broad and unclear requirements to pass the auditing process without changes in practice (Behnam & MacLean, 2011).
Although these three mechanisms foster adherence, some room for adaptation is advisable in the adoption phase as well, to address the risk of means-ends decoupling that is present, even if less prominent, in this phase. A potential counterbalancing adaptation mechanism is to create and maintain a space for dialogue and contestation, through which the recipients can express their concerns and frustration with some requirements or the barriers emerging to their implementation and ad hoc solutions can be designed jointly with the source of knowledge. Contestation is an essential aspect of VSS, as their design and implementation involve a variety of actors with often divergent views (Arenas, Albareda, & Goodman, 2020). Spaces to discuss emerging barriers to the implementation of requirements are useful for two reasons. First, such spaces allow the recipient to propose adaptations when the implementation of requirements is not feasible or their negative consequences are immediately visible, while giving the source of knowledge veto power on the proposed adaptations to limit the risk of opportunistic interpretation of the requirements. For instance, the International Sustainability and Carbon Certification (ISCC), a multistakeholder initiative sponsored by the European Commission that aims to ensure the sustainable sourcing of raw materials, regularly holds regional stakeholder dialogues, where the adopters, the standard-setter, auditors, and other stakeholders can discuss implementation challenges to encourage adaptation. Second, when the difficulties in initial implementation are shared and discussed with the source of knowledge, joint solutions can be developed. The source of knowledge can, for instance, provide training for specific needed skills or grant greater empowerment to the recipient if the high pressure for compliance is a source of demotivation (Egels-Zandén, 2014;Kim, 2013).

308
Business Ethics Quarterly

Mechanisms in the Integration Phase
Here again, we distinguish a set of mechanisms related to the source of knowledge, the recipient of knowledge, and the knowledge transferred, but, in this phase, with a focus on adaptation, given the prominent risk of means-ends decoupling.
Mirroring the knowledge transfer literature, in this phase, the role of the knowledge source becomes less prominent, while the role of the recipient is key to performing the requirements daily (Szulanski, 1996(Szulanski, , 2000. As the relevant knowledge shifts from knowing the VSS requirements to knowing how to implement them effectively in the local context, global sources of knowledge are likely to be substituted by local networks for knowledge creation and diffusion, such as the local networks for the UNGC or the Data Partners of the GRI. Local sources of knowledge (i.e., country-or region-level actors supporting VSS implementation, including local governmental actors, companies, and NGOs) help to increase the fit between the VSS and the context of implementation and to support a tailored implementation, as they are better positioned than global sources of knowledge to support the local adaptation of VSS requirements. They possess superior knowledge about the local context and reduce the burden on the recipient of knowledge to identify the most opportune ways to adapt the VSS requirements and to steer the certified organizations away from misguided adaptation. For instance, interactions with peers foster understanding, interpretation, experiential learning, and the diffusion of best practices, leading to an intelligent adaptation of the VSS requirements to the local context (Perez-Aleman, 2011). Likewise, Bolivian nut producers succeeded in meeting stringent European Union (EU) food safety standards because they were able to cooperate with each other to implement the VSS requirements (Coslovsky, 2014). In the dairy sector in Nepal, cooperatives performed best among the different possible coordination arrangements (Kumar et al., 2017).
As posited by the knowledge transfer literature, the knowledge recipients play a more crucial role in this second phase, as they need to integrate fully the new knowledge into daily routines (Szulanski & Jensen, 2006). We argue that a recipient of knowledge that is autonomous (i.e., that has less frequent exchanges and lower monitoring) from the source of knowledge favors effective adaptation and the routinization of the new practices in the integration phase for two reasons. Higher autonomy from the source of knowledge allows for participation in local networks, where locally adapted practices are developed and diffused. The risk of close relationships with global sources of knowledge is an overreliance on the existing transferred knowledge, which can undermine the creation of new local knowledge essential to overcoming means-ends decoupling. The recipients of knowledge have to exercise considerable agency and creativity to "wrangle" external requirements into the realities of their organizations, resisting the pressure from the source of knowledge to implement requirements "by the letter" by proposing adaptations that are functional to reach the VSS intended goals in the local context (Wijen, 2014). For example, some Brazilian cocoa producers have successfully integrated VSS requirements into local practices, not as a result of government or standard-setter pressure, but through the efforts of local adopters that have been experimenting autonomously 309 Implementing Voluntary Sustainability Standards with ways to meet global VSS through creating locally suitable practices in response to an international standard (Tampe, 2021).
Finally, the knowledge transfer literature posits that a certain degree of flexibility and interpretability increases the fit between the new knowledge and the local context, overcoming material and institutional barriers (Ansari et al., 2010). In the VSS context as well, when the knowledge transferred is interpretable, there is a lower risk of means-ends decoupling (Wijen, 2014). Interpretability refers to the possibility to implement a given requirement in multiple ways, differing in the process or outcome reached, while still being compliant. In this phase, the possibility to discuss diverging interpretations of the requirements can also stimulate the discussion on how to reach the final goal of the VSS effectively, given the context specificities. In line with our argument, Christensen and colleagues (2017) argue that the possibility to interpret VSS requirements in multiple ways is not a barrier but a catalyst for reaching the ultimate goals of VSS, because open-ended requirements trigger a discussion among different actors that sensitizes them to the goal, rather than to the letter of the requirements, and that allows them to adapt the requirements to maximize their effectiveness.
While these three mechanisms favor adaptation, reducing the risk of means-ends decoupling, they should be counterbalanced by maintaining adherence mechanisms, as the policy-practice decoupling risk persists in the integration phase. A potential counterbalancing mechanism is the implementation of a commitment-oriented approach by the knowledge source, based on collecting information about the recipient's state of implementation of requirements, not to punish noncompliance but to establish an ongoing conversation with the recipient. Such an approach allows for joint problem solving during the implementation and integration of requirements (Locke et al., 2009). Moreover, the source of knowledge can verify that the proposed adaptations are functional to boost VSS effectiveness in the local context, while the spirit of the VSS is preserved. For VSS schemes that involve periodic monitoring, the traditional monitoring based on sanctions for lack of compliance can be complemented with pedagogical monitoring that supports local innovativeness, instead of punishing recipients of knowledge for engaging in adaptations (Coslovsky, 2013;Josserand & Kaine, 2016;Locke, 2013;Locke et al., 2009). For instance, Coslovsky (2013) analyzes the case of a cooperative of sugar producers in Brazil that succeeded in a demanding environment because of strong collaboration between the auditors and the adopters. Auditors acted as conduits of information with the standard-setter and also between and within organizations, providing technical support for emerging internal problems, rather than as police officers who detected violations and imposed fines. For VSS that do not have a formal periodic monitoring mechanism, the commitment of the source of knowledge to monitor the recipient's implementation over the long term can be granted through ad hoc channels, such as periodic stakeholder dialogues, reviewed periodic reports, and regional hubs that, better than global headquarters, favor direct relationships and collaboration between the source and the recipient of knowledge over the long term (Pope & Lim, 2020).
In sum, in each phase, the actors involved in VSS implementation should favor the primary need (adherence and adaptation, respectively, in the two phases), while 310 Business Ethics Quarterly ensuring the presence of counterbalancing mechanisms that, albeit in the background, support the opposite need. In the next section, we unpack the tensions that exist between the mechanisms in the two phases along the transition from the adoption phase to the integration phase.

Tensions between Phases
When comparing the mechanisms required to overcome the prominent risks of policy-practice decoupling in the adoption phase and of means-ends decoupling in the integration phase, it becomes evident that each phase requires oppositional mechanisms. Adopting a dynamic perspective draws attention particularly to the transition between phases when the source and recipient of knowledge need to change how they enact VSS implementation and when the knowledge transferred benefits from a shift in its nature. In this transition from the adoption to the integration phase, three tensions emerge. First, a tension of proximity arises between global and local sources of knowledge. Global sources of knowledge are beneficial in the adoption phase, as they have greater motivation to monitor the correct implementation of requirements and increase the pressure for adherence (Donaghey et al., 2014;Schuler & Christmann, 2011). However, in the integration phase, global sources of knowledge should leave room for local sources that are better able to guide the recipients of knowledge in their adaptation efforts (Coslovsky, 2014;Perez-Aleman, 2011), while still retaining some control over the recipients' adaptations to prevent unfaithful enactment of the requirements.
Second, a tension of autonomy emerges because the adoption phase benefits from recipients close to the source of knowledge, whereas the integration phase benefits from recipients that are autonomous from the source, while still allowing for some external control from standard-setters to prevent devious adaptations. Close relationships with the source of knowledge increase the acceptance of requirements and the adherence to them (Josserand & Kaine, 2016;Locke, 2013;Pipkin & Fuentes, 2017), whereas local autonomy is beneficial in the integration phase to develop local knowledge and increase the fit of VSS with the local context (Tampe, 2021). The transition between the two phases is therefore particularly challenging for the recipient of knowledge, which needs to move the internal focus from accepting pressures for adhering to requirements to experimenting with bottom-up innovations.
Finally, a tension of interpretability occurs because clear and unambiguous transferred knowledge is beneficial for initial implementation during the adoption phase, whereas interpretable transferred knowledge supports the integration phase. Clear and unambiguous requirements are beneficial in the adoption phase to improve recipients' understanding and acceptance of requirements and to reduce opportunistic interpretations and policy-practice decoupling (Behnam & MacLean, 2011;Egels-Zandén, 2014;Giuliani et al., 2017), whereas interpretable requirements facilitate the fit between the VSS and the context of implementation in the integration phase, thus reducing the risk of means-ends decoupling (Ansari et al., 2010;Wijen, 2014). Yet, while taking a more prominent role, the interpretability of requirements should not fully undermine clarity in the integration phase, which is still needed to prevent the recipient of knowledge from an adaptation of requirements that is too far from the spirit of the VSS.
In sum, the adoption of a dynamic perspective reveals three tensions that emerge as the source of knowledge, the recipient of knowledge, and the transferred knowledge shift from a focus on adherence to a focus on adaptation. These tensions reveal an important pattern for VSS implementation that has remained underexplored in existing theory. However, it would be simplistic to assume that these tensions play out identically for different types of VSS. In the next section, we discuss the contingencies of the three tensions outlined based on VSS characteristics, adding further nuance to our framework.

APPLYING THE FRAMEWORK TO DIFFERENT VSS
In this section, we develop propositions to illustrate how our framework helps identify key implementation challenges for VSS with different characteristics, namely, stringency, enforcement (Potoski & Prakash, 2009), and scope (Auld, 2014;Reinecke et al., 2012). We propose that, whereas stringency enhances the tensions identified in our framework by placing emphasis on both poles of each tension, strong enforcement and broad scope pose greater emphasis on one pole, adherence or adaptation, resulting in less prominent tensions but in greater risk of decoupling and a need for counterbalancing mechanisms. Table 1 summarizes our arguments. On the basis of our propositions, we also show how combinations of these characteristics are likely to make the tensions underlined in our framework more or less prominent.

Stringency: Enhances All Tensions
Stringency is considered a key characteristic of VSS . VSS can be distinguished into stringent VSS that go well beyond the public regulations of the context of implementation and lenient VSS that require the same or  (Potoski & Prakash, 2009). Lenient VSS are therefore also less costly to implement, because the requirements are generally accepted principles that are more familiar to the adopter and need less knowledge to be understood and correctly implemented. An example is ISO 14001, which requires the implementation of a generic environmental management system, applicable to many contexts (Aravind & Christmann, 2011;King et al., 2005). Stringent VSS, in contrast, diffuse more ambitious sustainable practices and afford higher reputational benefits than lenient VSS. An example is the EU's Eco-Management and Audit Scheme, which targets organizations that have already achieved good environmental management performance, and have often obtained other certifications, such as ISO 14001, but want to take it a step further and commit to higher environmental performance. However, they are more costly and challenging to implement, unless an adopter with already advanced practices seeks to adopt the VSS to formalize practices already in place. A study that measured the implementation of different VSS requirements by agricultural producers in South Africa shows that lenient practices are more likely to be effectively implemented than stringent ones (Thorlakson, Hainmueller, & Lambin, 2018). We argue that the key challenge of stringent VSS implementation that affects both phases is that advanced knowledge needs to be transferred, which intensifies all three tensions. First, the source of knowledge is subject to a heightened tension of proximity. During the adoption phase, stringent VSS likely require the source of knowledge to provide significant guidance on the VSS implementation. Therefore stringency tends to increase the demand for global sources of knowledge that can monitor the knowledge implementation and, given their global exposure, provide significant reputational benefits, which are important to increase the acceptance of stringent VSS (Potoski & Prakash, 2009). Yet, during integration, the advanced knowledge inherent in stringent VSS requires a deep change in practices and increases the chance of misfit with the local context (Ansari et al., 2010), given that these VSS imply a significant distance between the VSS requirements and applicable public laws (Potoski & Prakash, 2009). This gap fosters high pressure for adaptation, calling for local sources of knowledge to develop and diffuse locally adapted practices (Pipkin & Fuentes, 2017).
Concomitantly, stringent VSS pose high demands on the recipients to implement the VSS, intensifying the tension of autonomy. Given the challenge of transferring advanced knowledge, stringent VSS may result in poor implementation because adopters are unwilling to live up to these demands (Dietz et al., 2020), requiring high adherence mechanisms. Thus, during the adoption phase, stringent VSS benefit from recipients close to the source of knowledge to fill the knowledge gap in the adoption phase. However, during integration, stringent VSS benefit from recipients autonomous from the source of knowledge to support the development of adapted practices. Room for local experimentation is especially relevant, because locally adapted practices are less likely to be diffused for stringent VSS, as they are generally less widespread than lenient VSS (Potoski & Prakash, 2009).
Finally, the advanced knowledge means that requirements go beyond national regulations and are generally less familiar to the recipient, which heightens the 313 Implementing Voluntary Sustainability Standards tension of interpretability. Given the recipients' lack of familiarity, stringent VSS tend to require clear and unambiguous knowledge transferred in the adoption phase to foster adherence. In contrast, in the integration phase, an interpretable transferred knowledge facilitates adapting the requirements to the context to overcome a lack of supportive industrial policies and a misfit between VSS requirements and local regulations (Distelhorst, Hainmueller, & Locke, 2017;Locke, 2013). Therefore we develop the following proposition: Proposition 1: The higher the stringency of a VSS is, the stronger is the emphasis on both poles of the tensions of proximity, autonomy, and interpretability, thus increasing the tensions' saliency.

Enforcement: Weak Emphasis on Adaptation
The second VSS characteristic proposed by Potoski and Prakash (2009) is enforcement. They distinguish three main enforcement mechanisms: independent thirdparty monitoring, public disclosure of audit information, and sanctioning of poor compliance. An example of a VSS with strict enforcement mechanisms is the ISCC certification for biomass and bioenergy. The ISCC certification scheme includes periodic audits by accredited third parties and occasional audits by ISCC, a public list of all certified organizations, and sanction mechanisms that can lead to the noncompliant organization being placed on a "blacklist" on the ISCC website and having its certification revoked for up to sixty months (ISCC, 2016). An example of the other extreme is the UNGC, which does not include enforcement mechanisms (Rasche & Waddock, 2014).
Strong enforcement accentuates the adherence pole of each tension by providing the adherence mechanisms identified in our framework as supporting the adoption phase. Strongly enforced VSS are generally employed in global supply chains, promoted by global sources of knowledge to guarantee to consumers in the industrialized world that certain sustainability goals are met also in countries with weak regulations (Potoski & Prakash, 2009). Moreover, as strongly enforced VSS require a formal process of evaluation of how the requirements are implemented, the source of knowledge interacts, directly or through auditors, with the recipient of knowledge to award the certification and to monitor the implementation, resulting in recipients close to the source of knowledge. Finally, as the implementation of requirements needs to be easily assessed and monitored, strongly enforced VSS tend to have clear and unambiguous transferred knowledge (Egels-Zandén, 2014; Giuliani et al., 2017).
However, the key challenge for strongly enforced VSS is the weak emphasis on adaptation. The preponderance of adherence mechanisms comes at the expense of counterbalancing mechanisms for adaptation in the adoption phase. The focus on adherence is likely to make the recipients of knowledge more resistant, in line with the proposition that using force for effecting change in someone else's behaviors is associated with a low level of stability of the intended behavior (Lawrence, Winn, & Jennings, 2001). Therefore, in the adoption phase, a space for dialogue and contestation helps attenuate the strong focus on adherence (Arenas et al., 2020;Soundararajan, Brown, & Wicks, 2019).

314
Business Ethics Quarterly The tendency against adaptation persists in the integration phase, where the weak emphasis on the three adaptation mechanisms identified in our framework leads to an increased risk of means-ends decoupling. When strong enforcement mechanisms are in place, adaptation has a high cost and risk for the adopter: the implemented adaptations can be perceived as noncompliant by the auditors and punished accordingly, reducing the role of local sources of knowledge in diffusing adaptation and increasing the costs for recipients' autonomy (Wijen, 2014). Moreover, to facilitate monitoring and punishment for noncompliance, strongly enforced VSS are less likely than weakly enforced VSS to present interpretable transferred knowledge that favors local adaptation. In other words, strongly enforced VSS do not give prominence to local sources of knowledge, autonomous recipients, or interpretable knowledge, which, as illustrated in our framework, facilitate the integration phase and reduce the risk of means-ends decoupling. As a result, we develop the following proposition: Proposition 2: The stronger the enforcement of a VSS is, the stronger is the emphasis on the adherence poles of the tensions of proximity, autonomy, and interpretability, thus increasing the need for counterbalancing adaptation mechanisms in the adoption phase and the risk of means-ends decoupling in the integration phase.

Broad Scope: Weak Emphasis on Adherence
The third VSS characteristic relevant to apply to our framework is the scope of VSS. Broad-scope VSS cover a variety of issues (Auld, 2014) and are widespread across different types of organizations (Reinecke et al., 2012), geographies, and industries ). An illustration is the aforementioned ISO 14001, with more than 348,000 certificates in 176 countries (International Organization for Standardization [ISO], 2020), or Rainforest Alliance, a VSS covering social and environmental issues across 70 countries and multiple crops. In contrast, VSS with a narrow scope target specific sectors, such as textiles (e.g., codes of conduct of the Fair Labor Association or Worker Rights Consortium), electronics (e.g., the Responsible Business Alliance Code of Conduct), or soy (e.g., the Round Table on Sustainable Soy); organizations, such as small agricultural producers (Reinecke et al., 2012); or regions, such as a Nicaraguan "Trademark of Trust" label for local agricultural producers (Starobin, 2021).
The key challenge for implementing broad-scope VSS, we argue, is the weak emphasis on adherence mechanisms, which increases the risk of policy-practice decoupling in the adoption phase. First, as broad-scope VSS are widespread and adopted in various sectors and geographic areas, the standard-setter acts often indirectly via third parties that support adopters' implementation, as in the case of ISO 14001, limiting the role of global sources of knowledge. Second, the mediation via third parties results in recipients of knowledge being less close to the source of knowledge compared to what is found with equally enforced but less broad VSS. Third, these VSS are less likely to present clear and unambiguous transferred knowledge, as the same requirements need to be implemented, for instance, by a 315 Implementing Voluntary Sustainability Standards mining company in Canada and a garment factory in Bangladesh (ISO, 2020). The more the VSS aspire to be diffused across different organizations and contexts, the greater is the incentive to create broad and general guidelines that can be implemented in multiple ways, as in the case of the UNGC or ISO 14001 (Haack & Rasche, 2021;Pope & Lim, 2020). The opposite extreme is represented by company-led VSS, adopted by suppliers of a large company, which typically include specific requirements for the production of a particular commodity and which support adopters through training and strict monitoring .
In contrast, during integration, broad-scope VSS are likely to provide adequate mechanisms for adaptation, as local sources of knowledge have a prominent role to develop and distribute a locally specific version of the requirements and of how to implement them. For instance, both the UNGC and the FLO established local networks able to engage with adopters through contextualized discussions (Auld, 2014;Rasche, 2012). Moreover, given the greater distance between the standardsetter and the adopters, broad-scope VSS are likely to grant autonomy to the recipients of knowledge to adapt the requirements to the respective context (Manning & Reinecke, 2016). Finally, adaptation is facilitated by interpretable knowledge, which supports the diffusion and applicability across different contexts. For instance, the ten principles promoted by the UNGC are kept abstract and broad to fit a variety of contexts and organizations (Rasche & Waddock, 2014).
However, such flexibility comes at the expense of counterbalancing adherence mechanisms, and specifically a commitment-oriented approach of the source of knowledge, that can prevent adaptation from going too far from the VSS intended goal. Empirical evidence comes from ISO 14001, where monitoring is entrusted to local auditors that tend to have looser ties with the source of knowledge, the ISO, than with the adopters and where studies have raised concerns about auditors' independence and potential conflicts of interest (Aravind & Christmann, 2011;Christmann & Taylor, 2006). Overall, broad-scope VSS tend to emphasize the adaptation poles of each tension, which is likely to increase the risk of policypractice decoupling and the need for counterbalancing adherence mechanisms. Accordingly, we develop the following proposition: Proposition 3: The broader the scope of a VSS is, the stronger is the emphasis on the adaptation poles of the tensions of proximity, autonomy, and interpretability, thus increasing the risk of policy-practice decoupling in the adoption phase and the need for counterbalancing adherence mechanisms in the integration phase.
Importantly, stringency and leniency, strong and weak enforcement, and narrow and broad scope should not be understood as binary categorizations but rather as a continuum between these extremes. Moreover, these characteristics are not present in isolation; rather, each VSS is characterized by a degree of stringency and enforcement and a more or less broad scope. Therefore, to understand how tensions emerge and evolve in the implementation of each VSS, it is necessary to consider the combinations of these characteristics and their effects on the framework we propose.

316
Business Ethics Quarterly Next, we discuss which combinations of stringency, enforcement, and scope are most likely to enhance or reduce tensions in implementation.

Combinations of VSS Characteristics
We expect three combinations-stringency and strong enforcement, stringency and broad scope, and strong enforcement and broad scope-to increase the three tensions, because the first two are further intensifying the already salient tensions for stringency, whereas the third one combines opposing poles of the tensions. Stringency and strong enforcement simultaneously increase the pressure for adherence and adaptation because stringency heightens the pressure for adaptation mechanisms that strongly enforced VSS are less likely to provide. For stringency and broad scope, stringency intensifies the pressure for adherence mechanisms, which are less likely to be present for broad-scope VSS. Finally, combining strong enforcement and broad scope is similarly problematic, because the former emphasizes adherence mechanisms, whereas the latter strengthens adaptation mechanisms, making the transition between the two phases difficult. An illustrative example of the first combination is Bonsucro, a stringent and strongly enforced VSS aiming at diffusing sustainable practices among sugarcane producers in the Global South. The standardsetter realized that the standard was adopted mainly by already advanced producers due to implementation challenges attributable to the lack of fit between requirements and local contexts (Thorpe, Guijt, Sprenger, & Stibbe, 2021), in line with our argument that stringent, strongly enforced VSS are less likely to provide the adaptation mechanisms required to implement stringent VSS. To reduce the tensions and generate a wider transformation in the sector, Bonsucro has recently included "adaptability" as one of its core principles, reached through the empowering of community leaders and the development of location-specific objectives that allow for contextual adaptation (Bonsucro, 2021), while also ensuring training and the creation of a supportive environment for more producers to transfer the relevant knowledge and ensure adherence (Thorpe et al., 2021). On the contrary, our framework suggests less prominent tensions for four combinations-leniency and strong enforcement, leniency and broad scope, strong enforcement and narrow scope, and weak enforcement and broad scope. The rationale for these combinations is that leniency reduces the pressure for adherence of strong enforcement and the pressure for adaptation of broad scope. Similarly, strong enforcement with a narrow scope, such as found with the company-led Nespresso AAA, or weak enforcement with a broad scope, such as found with the UNGC, likely attenuates tensions along the implementation, as these combinations, again, tend to reduce the pressure for adherence or adaptation, respectively. For example, in the case of the UNGC, the pressure for adaptation coming from its broad scope is reduced by the leniency of its ten universally accepted principles, while the lack of enforcement grants significant autonomy to the recipients throughout the process, allowing for interpretation of the rules and for the support of local networks.
Other combinations are possible. However, for these, we do not expect particularly enhanced or reduced tensions based on the propositions ensuing from our framework.

Implementing Voluntary Sustainability Standards
To summarize, applying our framework sheds light on what characteristics and combinations thereof are more or less likely to foreground or background the tensions and thus deepens our understanding of VSS implementation.

DISCUSSION AND CONCLUSION
VSS, despite their potential, have not lived up to the promises of better social and environmental performance. In light of the complex track record of VSS implementation, scholars have offered two alternative explanations, namely, policy-practice decoupling and means-ends decoupling, calling, respectively, for higher adherence or higher adaptation of requirements and pointing out the tension between the two (Bromley & Powell, 2012;Wijen, 2014). Building on these explanations and theorizing a dynamic perspective on the implementation of VSS and related tensions, we make three contributions.
First, while the VSS literature has recognized difficulties in the implementation of VSS because of a tension between adherence and adaptation (Brunsson et al., 2012;de Bakker et al., 2019;Wijen, 2014), our framework further emphasizes the importance of adopting a dynamic and temporal approach to this tension, illustrating its evolution over the course of VSS implementation. Thanks to the adoption of a knowledge transfer lens, we move beyond the dominant view of VSS implementation as an act and add theoretical foundations to studies with a longitudinal perspective on VSS implementation (Egels-Zandén, 2014; Lazaric & Denis, 2005;Sandholtz, 2012). Specifically, our article theorizes different phases of VSS implementation and discusses their differential needs for adherence and adaptation, overcoming the static view dominant in the extant literature that would suggest a stable focus either on adherence (Aravind & Christmann, 2011), on adaptation (Wijen, 2014), or on balancing these two needs without taking into account different phases of implementation (de Bakker et al., 2019;Mena & Palazzo, 2012;Overdevest & Zeitlin, 2014). Our framework also shows how three different kinds of tensions manifest particularly in the transition from the adoption to the integration phase, when the emphasis needs to shift from adherence to adaptation pressures. Bringing out these tensions and contradictory pressures between phases adds to our understanding of why VSS implementation has been so difficult to achieve in practice (Giuliani et al., 2017;Locke et al., 2009;Tolentino-Zondervan et al., 2016).
Second, our framework provides additional insights into how tensions between adherence and adaptation manifest based on different VSS characteristics, namely, stringency, enforcement, and scope. We also present a more fine-grained explanation for the lack of effectiveness of stringent VSS emerging empirically , especially when combined with strong enforcement or broad scope, as well as the lack of effectiveness of strongly enforced VSS with a broad scope (Aravind & Christmann, 2011;Christmann & Taylor, 2006). These combinations generate particularly strong tensions in the transition between the adoption and integration phases. This theoretical expectation is in line with empirical research on how actors deal effectively with competing needs and tensions, namely, that they need to embrace them, adopt a both/and rather than an either/or stance, and pursue a 318 Business Ethics Quarterly balance between these conflicting yet coexisting needs through enacting mechanisms that address the competing needs (D'Adderio, 2014). Third, we contribute to the literature on decoupling in VSS (Aravind & Christmann, 2011;Behnam & MacLean, 2011;Brunsson et al., 2012;King et al., 2005) by offering an explanation of how different kinds of decoupling intertwine and evolve throughout VSS implementation. Both kinds of decoupling play a role within phases, but with different importance: the focus on adherence to overcome the prominent risk of policy-practice decoupling during the adoption phase requires counterbalancing adaptation mechanisms as well. Similarly, in the integration phase of implementation, a focus on adaptation is needed to overcome means-ends decoupling, and yet adherence mechanisms should be present, too, to counterbalance adaptation. Moreover, our framework shows how decoupling of VSS and outcomes is not merely the result of lack of adherence to the requirements (Aravind & Christmann, 2011;Christmann & Taylor, 2006;Egels-Zandén, 2014) nor lack of adaptation to local specificities (Christensen et al., 2017;Rasche, 2010;Sandholtz, 2012;Wijen, 2014); it can also arise from the difficulties in reaching and maintaining the balance between the contradictory needs of adherence and adaptation dynamically over time.
Three boundary conditions are worth noting, calling for further refinement and extension of our framework to incorporate additional dimensions. First, we are aware that the knowledge transfer lens that we bring to the analysis emphasizes the cognitive dimension of VSS implementation. It considers the dynamics inherent in new knowledge implementation and capability building but backgrounds other important aspects of VSS implementation, such as political, cultural, or material aspects. Our framework focuses on how the pressure for adherence and adaptation affects the implementation of VSS once adopted, without accounting for their effects on the decision to adopt a particular VSS. Moreover, we rely on the assumption that the adherence-adaptation tension faced in VSS implementation is parallel to the one faced during knowledge transfer within or across organizations. While this assumption seems plausible, further empirical research, paying particular attention to issues of power and voice between actors that characterize the VSS context, is needed to examine and possibly refine this assumption.
Second, the counterbalancing mechanisms proposed in the two phases are not intended as a panacea. Both mechanisms call for closer interaction between the source and the recipient of knowledge. However, power dynamics, lack of trust or understanding, or too-heated contestation might undermine, instead of enabling, the implementation of VSS. The capability to counterbalance the prominent need depends on the capacity of the parties to collaborate and negotiate. When collaboration is difficult to achieve and relationships are arduous, the proposed mechanisms might instead slow down or even undermine the implementation of VSS. In short, these solutions are often more apt to "govern the gap" between the VSS and reality, rather than eliminating the gap (Huising & Silbey, 2011).
Finally, in applying our framework, we simplify manifold VSS initiatives based on three characteristics, namely, stringency, enforcement, and scope Potoski & Prakash, 2009), to discuss the saliency of the tensions 319 Implementing Voluntary Sustainability Standards emerging from our framework. However, VSS can be categorized along other characteristics that are orthogonal to the ones we use. For example, VSS can be distinguished based on the number of actors involved in the VSS governance (de Bakker et al., 2019), the specific content of the VSS , or their orientation toward processes or outcomes (Brunsson et al., 2012). Therefore further theorization is required for additional characteristics.
We encourage future research to empirically test and refine our insights into how the tensions between adherence and adaptation evolve over time and how they might shape VSS implementation strategies, for example, by adopting a process perspective through longitudinal case studies. The goal is to analyze further the two phases of VSS implementation delineated in our framework and to shed light on the implementation challenges and differential decoupling risks affecting each phase, ideally with a comparative approach considering different VSS and contexts (Langley, 1999;Langley, Smallman, & Tsoukas, 2013). Given our expectation that stringent VSS are particularly prone to tensions, we consider these kinds of VSS as well suited to empirically testing our framework.
Moreover, our framework speaks to decoupling as an impediment to VSS implementation and reveals new insights into tensions that arise from the contradictory needs to enforce adherence and to encourage adaptation. A follow-up question is how the actors involved in VSS implementation could recouple VSS requirements and outcomes, considering the temporal evolution of the tensions between adherence and adaptation. Future research should explore how the actors involved can overcome the tension of proximity and ensure VSS global exposure without neglecting the presence of local networks with a more decisive role during the integration phase. Similarly, we encourage examining strategies that reduce the tension of autonomy by fostering initial closeness between the standard-setter and the adopting organization, without impeding the latter's future autonomy. Finally, future research can explore how standard-setters can reduce the tension of interpretability by designing requirements that become more interpretable over time. Ethnographic approaches and case studies can dive into the microlevel and governance challenges that organizations are likely to experience as they reconcile competing needs in the transition between the two phases, while quantitative research comparing different VSS can analyze which strategies are most effective for recoupling VSS and outcomes.
To conclude, VSS remain an important pathway toward a more sustainable future despite implementation challenges. As VSS initiatives have high opportunity costs for adopters, it behooves scholars and practitioners alike to minimize such failed undertakings. This article highlights different kinds of tensions between adherence and adaptation as key obstacles to the successful implementation of different types of VSS and unpacks these tensions by adopting a dynamic perspective. A key task for VSS actors is to manage tensions effectively, considering differing adherence and adaptation needs and navigating the transition between the conflicting yet interrelated demands of each phase. We hope that our theorizing on VSS implementation dynamics will inspire further conceptual and empirical work on how 320 Business Ethics Quarterly actors can do so, because developing a deeper understanding of ways to address tensions is essential for VSS to live up to their promise of fostering ethical behaviors in practice.