Do companies walk the talk? Commitments and actions in global supply chain labor standards

Abstract

This article examines the efforts towards implementing minimum labor standards in global supply chains through the lens of corporate social responsibility. The adoption of supplier codes of conduct has driven efforts to monitor and enforce standards within these chains. Nonetheless, challenges persist in translating commitment into action, giving rise to gaps in implementation. We address two critical phenomena: corporations' varying internalization of responsibility to implement labor standards, and the translation of these written commitments into actions. Through two distinct studies, we explore how companies ‘walk the talk’, and navigate the gaps between corporate recognition of responsibility and the establishment of management systems to implement labor standards. The first study demonstrates that a minority of companies are diligent in committing to implement their codes. Only 17% accept a shared responsibility to implement the code, and most companies only refer to audit visits or the termination of supplier contract as mechanisms put in place. The second study does not find conclusive evidence of a correlation between corporate commitments and the evaluation of their corporate sustainability practices in supply chains as evaluated by KnowtheChain. However, results suggest that companies do better when they promote the collaborative approach with suppliers in their implementation efforts.

Introduction

Private regulatory efforts to address labor and environmental risks in global supply chains led to the widespread adoption of supplier codes of conduct (SCCs), most of them including minimum labor standards applicable to supply chain workers. As highlighted by Vandenbroucke et al. (2024), over 80% of codes refer to child labor, forced labor, collective bargaining and freedom of association, and discrimination. These commitments mark the growing concern of private actors for non-financial matters. Awareness on human rights and environmental issues in global supply chains increasingly translates into corporations’ strategies and policies, forming the so-called corporate self-regulation and a private governance around these issues.¹ However, the narrative of the ‘sustainable’ or ‘ethical’ company is controverted, and many question the actions undertaken by companies, beyond their words and policy standards proudly published for the world – especially the consumer – to see.

According to international soft law, it is expected that firms act upon their standards dictated in their codes and ensure implementation. For instance, the United Nations Guiding Principles on Business and Human Rights (UNGP) principles #19 and #20 emphasize the importance of tracking the effectiveness of actions taken to prevent and mitigate adverse human rights impacts. The OECD Guidelines for Multinationals Enterprises suggest several measures that companies should take to ensure minimum labor standards protection, including supplier assessment and conducting due diligence to ensure compliance with codes, collaborating and engaging with suppliers to address challenges, and establishing procedures for addressing non-compliance, amongst other things.

Despite these international recommendations, previous studies assessing implementation efforts of corporate self-regulation highlight the challenges of turning words into actions (i.e., policy standards into decent labor conditions) in the global supply chain. Mamic’s research in 2005 reported the emergence of implementation programs within corporations, including audit practices, reporting procedures, training initiatives, remediation mechanisms, and the pivotal role of stakeholder dialogue (Mamic 2005). Subsequent scholarly discussions on compliance programs point out the ineffectiveness of these implementation programs, their inherent challenges, or the issues in how they are executed (e.g., Coslovsky & Locke 2013; Locke & Samel 2018; Paiement 2021).

Today, the empirical literature gives us insights into which implementation mechanisms are the most effective to improve labor standards in the global supply chain, and which instruments should be included in supplier codes of conduct (see the literature review by Vandenbroucke 2024). However, we are missing data from corporate practices to fully understand the extent to which they internalize these expectations by integrating implementation provisions in SCCs; and put in place implementation programs in practice. Given the voluntary and largely unregulated nature of private governance of labor standards within the global supply chain, the development of implementation programs essentially relies on corporate willingness to embrace a sense of responsibility for their impacts. This article aims to provide quantitative data on implementation provisions and programs adopted by multinationals, shedding light on how companies currently ensure the effectiveness of their codes and contribute to the implementation process itself. We pose the following question: Which role do implementation programs as set up by multinationals play in limiting social risks in the global supply chain? This paper presents one of the first empirical studies that (1) provides descriptive data on the content of supplier codes of conduct regarding implementation provisions, (2) analyzes the quality of implementation provisions, and (3) compares written commitments with independent benchmark reports among a subset of companies in the database.

The first section reviews the literature to map conceptual approaches on corporate responsibility to limit social risks in the global supply chain. The second and third section present empirical results from two studies that we conducted. The first study investigates implementation provisions laid down in SCCs by assessing the textual content of codes. The second study investigates the gap between the written provisions in supplier codes of conduct and the existence of programs in practice. We do so by comparing our own database of SCCs with the database built by KnowtheChain. We offer insights into how SCC textual content corresponds with corporate actions implementing their standards, thus how companies ‘walk the talk’.²

Conceptual framework: multinational responsibility to implement labor standards

Within the field of global governance, theoretical discussions on corporate responsibility are developed from different angles. Political philosophy theorists discuss the notion of shared responsibility, to understand the role of multinationals in their global supply chains beyond a liability model. From the perspective of corporations, stakeholder theories help us understand the shift from the profit-driven firm to the social enterprise that society expects, and how companies embrace this responsibility. In practice, institutional theories tell us how decoupling can occur between the words and actions of corporate social responsibility (CSR) strategies. Conceptualizing the terms ‘implementation provisions’ and ‘implementation programs’ is necessary to assess how corporate commitments translate into practice.

Theorizing shared responsibility beyond the liability model

The common way of ascribing responsibility in legal proceedings uses the liability model, when agents are held responsible for a specific outcome based on their direct causal contribution to it. Based on the liability model, blame and guilt are attributed to agents by taking into account the “fairly direct interaction between the wrongdoer and the wronged party” (Young, 2006a, p. 118). The concept of liability is central to legal reasoning, as it aims to attribute responsibility in a cause-effect relationship that is logic and relatively measurable. Political philosophy theorists have argued that this model of responsibility cannot be applied to global supply chain structures and injustices; interactions in the global market are too complex to reconstruct causal chains, the responsibility for human and environmental risks is shared among different actors, and global supply chains are difficult to trace (Hahn 2009). As a result, these structural injustices are the outcome of multiple actions and processes carried out by diverse agents, where a direct causal effect is hardly ever evident, and attribution of blame is not a mathematical formula.

As an alternative to the liability model, political theorists – with Iris Marion Young as their pioneer – propose the notion of shared responsibility, with the development of the social connection model of responsibility. This model argues that all individuals connected to structural injustice share a responsibility to collectively struggle against it, even when they are not directly responsible for causing them. She calls this a political responsibility (Young 2004). The social connection model has two main added value compared to the liability model, particularly relevant in the global supply chain context. Firstly, it emphasizes forward-looking responsibility, while the liability model which is backward looking and seeks responsibility for past actions. Secondly, in this model, actors jointly bear responsibility for the elimination of structural injustices, with the idea that the structure that creates injustices can only be changed if many agents engage in cooperative efforts. Applying Young’s theory, Zwolinski (2012) and Phillips (2022) conclude that responsibility does not only occur in case of wrongful participation in unjust structures. Even actors “minding their own businesses” and acting within accepted norms and rules participate in reproducing structural injustices.

Young’s social connection model forms a good theoretical basis for a delegation of responsibility for global justice to transnational private actors, beyond legal obligations and even when companies act within accepted norms and rules. While companies are not (yet) legally liable for their transnational social impact, they have a clear role as powerful actors in the global supply chains production systems. Based on these theories, many scholars focusing on corporate responsibilities toward wider society have been vocal about firms’ responsibility towards global society. For instance, Loosemore and Phua (2010) mention that contemporary business demands encompass addressing broader societal challenges, and that products and services are required to fulfill wider needs than before. Berkey (2021) contends that firms have extensive positive duties to the global poor, meaning they have an active role in preventing exploitation of employment and use the resources at their disposal to provide decent labor conditions.

Aßländer (2020) explores the concept of subsidiarity to allocate responsibility to the actors in global supply chain. As a governance principle, subsidiarity constitutes a priority rule that gives smaller entities precedence over higher instances. Only in cases where individual actors are unable to solve problems themselves, higher instances have a responsibility to assist. Aßländer considers that, as secondary actors in society, corporations bear moral obligations to ameliorate social and environmental conditions. Drawing inspiration from the Kantian perspective, the theorist argues that the abilities of the lower instances in society create a moral right for assistance for the lower‐level instances in society and simultaneously a perfect duty of assistance for higher‐order instances. In this ‘layered’ society of actors, he suggests that corporations can be seen as the ‘higher-order’ instance within the buyer–supplier top-down relationship. While this notion has faced criticism, Aßländer’s subsidiarity principle provides for an interesting perspective on the degree of corporate responsibility. When suppliers and workers struggle to implement labor standards, multinational enterprises (MNEs) have a moral obligation to step up and fulfill a role of assistantship towards the disadvantaged group of persons. In this sense, corporations bear duties to collaborate with others to overcome the problems at hand. To cite Aßländer’s (2020, p. 729) view: “based on the principle of subsidiarity, [we see this responsibility] as a responsibility which follows a hierarchical order shifting from primary actors to secondary actors in society if the respective tasks cannot be accomplished at the lower level”.

Theories on shared responsibility and subsidiarity explain why company are responsible for their global supply chains labor standards, even though they are not the direct perpetrator of wrongful labor conditions. Now, one may question whether companies accept this role and consider itself responsible. We investigate this in Study 1.

Embracing the paradigm of corporate responsibility

From the economic perspective, theories of capitalism and profit-driven enterprises have long guided the principles of business conduct. Milton Friedman's (1980) assertion that globalization is a 'win–win' scenario has echoed through boardrooms, advocating for the pursuit of profits as the primary goal of corporations. To integrate human rights into their business practices requires a shift redefining the raison d’être of corporations, from the singular profit-driven objective to the internalization of sustainability duties and responsibilities. The rising concept of "Creating Shared Value," as put forth by Porter and Kramer (2011), underscores the idea that societal needs and economic value are interconnected.

From the business perspective, stakeholder theory posits that companies should not merely mitigate harm but have positive impacts, fostering economic growth while addressing pressing global challenges. As underlined by Dmytriyev et al. (2021), stakeholder theory argues in favor of building stakeholder relationships and creating value for all stakeholders, thus including financial and non-financial elements. Stakeholder theory thus argue that the corporate landscape must embrace a new paradigm. A growing realization of the interconnectedness between business operations, the environment, and social well-being is reshaping the way companies perceive their roles and responsibilities. The popular adoption of SCCs by multinationals is one of the steps forward taken by corporations to signal to their stakeholders their awareness and attention towards issues beyond financial aspects. The CSR literature underlines an acknowledgement of businesses’ responsibility towards society, as CSR is considered good corporate behavior going beyond the core duties of a company through voluntary perspectives (Kolk 2010). Yet, as Tamvada (2023) explains, there is a lack of understanding and convergence on why, to what, and how corporates are responsible to society.

While the literature underlines that there is a shift, transforming corporations’ raison d’être, it is uncertain how (and if) this new role translates into practice at business level. Study 2 of this paper aims to give some elements of response to fill this gap.

Risks of decoupling: from policy to outcome

One of the risks of corporate self-regulation highlighted by scholars is to fall in the ‘window dressing’ trend, where standards formulated do not match the reality of business practices (Cerchia & Piccolo 2019). This leads to the decoupling between the formal policy and the actual practices, where a policy is formally introduced but not actually implemented in daily practice. This concept stems from institutional theories, developed by Bromley and Powell (2012) and later adapted to corporate practices in global supply chains, to explain the gaps between SCCs and their compliance (Bartley & Egels-Zandén 2015; Bird et al. 2019). As detailed in Vandenbroucke (2024), from the adoption of a supplier code of conduct to reaching its intended outcome, the process can be interrupted (decoupled) at three stages: there can be a goal-system decoupling, policy-practice decoupling, and a means-end decoupling.

In this paper, we investigate in two subsequent studies two decoupling stages: the goal-system decoupling and the policy-practice decoupling. Jointly, our studies examine whether corporations commit to taking an active role in addressing labor risks in their global supply chain, thus accepting their shared responsibility in reforming unjust structures. To this end, we delve into the complexities surrounding the implementation of labor standards in global supply chain and seek to assess the gaps between words and actions in global supply chains and understand corporate actions beyond the mere standard-setting goal of SCCs. As highlighted in the conceptual model (see Fig. 1), Study 1 investigates the goal-system decoupling by analyzing how corporations set goals as well as management systems to reach those goals in their corporate policies. A central point of focus here is to assess whether companies commit to participate in the implementation of labor standards when setting their management systems. Study 2 looks into the policy-practice decoupling by investigating whether internalization and recognition of responsibility for implementation by companies leads to effective actions and implementation programs.

Fig. 1

Conceptual model: Intentions of studies 1 and 2 to assess SCC decoupling process. Note. Conceptual model adapted to codes of conduct, from the theory by Bromley and Powell (2012) and readjusted by De Bree and Stoopendale (2020)

Conceptualizing implementation provisions and implementation programs

In line with the decoupling stages, we distinguish two concepts, implementation provisions and implementation practices. Implementation provisions are the written commitments laid down in the SCC, relevant at the goal-system decoupling phase. Implementation practices are the actions effectively undertaken by MNEs, appearing at the policy-practice decoupling stage. To implement labor standards, companies establish management systems. As defined by De Bree and Stoopendale (2020), management systems include all the kinds of intended organizational measures and procedures to achieve the goals set by the policy. This article studies the management systems designed to assess, facilitate and ensure compliance with labor standards stipulated in those codes, investigating those systems both as described in implementation provisions and as established in implementation practices.

International soft law and guidelines have previously indicated what type of management systems are recommended for an effective compliance with labor standards. According to the UNGPs, the ILO Tripartite Declaration of Principles Concerning Multinational Enterprises and Social Policy and the OECD Guidelines for Multinational Enterprises, business enterprises should carry out human rights due diligence. This process includes assessing actual and potential human rights impacts, integrating and acting upon the findings, tracking responses, and communicating how impacts are addressed. These texts also point out the importance of collaborating and consulting stakeholders to address systemic challenges and promote responsible business conduct. In case of SCC violation, the UNGPs detail the process to guarantee access to remedy for victims of labor rights violations in global supply chains. Moreover, these soft law instruments consider that multinationals’ responsibility to apply minimum labor standards extends beyond simply creating a policy document; it involves ongoing efforts to assess, address, and mitigate potential adverse impacts in supply chains. In our terms, management systems integrated in implementation should thus match implementation practices.

Study 1. Written commitments in implementation provisions

In this first study, we unravel the content of SCCs’ implementation provisions. Using two methods of analysis of implementation provisions, we measure the extent to which MNEs are invested not only in the standard setting of their self-regulatory policies, but more precisely in the establishment of implementation mechanisms. The methods employed in this study to analyze SCC content allow us to gain insights into MNEs’ investment to implement their codes – beyond the mere citing of labor standards.

Methods for study 1

Phase 1: Collecting supplier codes of conduct

The collection of supplier codes of conduct analyzed in this study was performed between September 2020 and June 2021, and contributed to the Database of Business Ethics (DBBE). The paper “Decoding supplier codes of conduct with content and text as data approaches” (Vandenbroucke et al. 2024) was the first published paper using the data from the DBBE. More information on the process of the data collection and the description of the methodology can be found there. This database gathers SCCs from a target sample of 1241 companies in 30 countries and counts 880 codes of conduct applying to suppliers. For this study, we have reduced this sample to 810 SCCs, by excluding all codes that applied both to employees of the company and the suppliers. It was important to focus on codes applying exclusively to suppliers, to avoid misunderstandings on the actors targeted by implementation mechanisms.

Phase 2: Measuring implementation provisions

To operationalize implementation provisions, we extracted paragraphs of SCCs pertaining to implementation and coded them to obtain measurable and quantifiable data. All content of the code fitting in the above-mentioned definition of implementation provisions (i.e., all management systems designed to assess, facilitate and ensure compliance with codes’ requirements) was extracted. This data extraction was performed manually between December 2022 and February 2023.

Phase 3: Text analysis using keywords method

We proceeded to analyze the extracted content with two different types of text analysis approaches (phases 3 and 4, respectively). Firstly, we used the dictionary method to capture the frequency of reference to selected terms (keywords). With this method, we intended to investigate corporate commitments to setting management systems. To identify the relevant keywords, we began by analyzing the Responsible Business Alliance (RBA)³ SCC, as this code represents a looked-up practice in business conduct for the drafting of SCCs. Within our database, over 80 companies have adopted the RBA code. Moreover, the RBA code includes a section dedicated to “Management Systems”, insisting on the following actions to ensure implementation: company commitment; management accountability and responsibility; legal and customer requirements; risk assessment and risk management, improvement objectives, training, communication, worker feedback, participation and grievance, audit assessments, corrective action process, documentation, and records, and finally supplier responsibility. Among these practices, we identified five categories of management systems, namely: transparency of the supply chain, risk and assessment monitoring, training programs, corrective action process, and reporting mechanisms. After a trial-and-error pilot of codes, we selected a list of keywords for each of these categories, laid down in Table 1. Presence of these keywords in SCC implementation paragraphs manifest a high likelihood that the SCC included this management system in their SCCs, meaning that they intend to establish this practice. For this method, the variable “implementation categories score” referred to the number of categories referred to in a SCC, regardless of the frequency of the appearing keywords. Regarding category 5, namely the corrective action process, it is noteworthy that we include “termina*” among the keywords, meaning that we consider the termination of the supplier–buyer contract as one of the management systems established to ensure SCC compliance.

Table 1 Definition and keywords of management system categories

Phase 4: Text analysis by expert assessment

Our second way of analyzing implementation provisions in SCCs was to ask experts to provide for a quality assessment of companies’ commitments to implementation provisions based on a group classification ranging from 0 (companies have no SCC) to 5 (best-case scenario of implementation provisions). This expert assessment was performed manually, with three different researchers doing the coding to ensure reliability of our classification and avoid biases. This classification is not only based on keywords, but the global meaning of the implementation provisions paragraph extracted, including its grammar. This research design was notably built on the findings by Crilly et al. (2016), according to whom the focus on the content or the grammar of how social actors explain and rationalize their conduct leads to different outcomes. While corporate actors may persuade stakeholders of their ethical behavior based on the content of discourse, its grammar reveals information on the intention of the company (Crilly et al. 2016). This manual method of text assessment aims to analyze the text beyond its content or mechanisms listed and look at how and why the provisions are phrased to understand the level of corporate commitment.

After reading more than 100 implementation provisions in SCCs, the experts discussed which categories would be an appropriate operationalization of the quality of implementation provisions. The five categories represent different levels of responsibility, imposed on suppliers (recommendation or obligation), and/or on the multinational itself. As we are trying to assess “To what extent are multinationals taking an active role in limiting social risks in their global supply chain via the establishment of implementation programs?”, the distribution of responsibility for implementation is relevant. Categories 3 and 4 mark the difference between the responsibility solely relying on suppliers to comply with the standards, and a sense of responsibility shared with the buyer itself present in category 4.

• 0. The company has not drafted a supplier code of conduct (this applies to 431 companies out of our initial database of 1241 companies, thus leaving us with a sample of 810 SCCs)

• 1. The SCC does not have any implementation provision, or not noteworthy.

• 2. Recommendation and encouragement to suppliers: Implementation provisions are formulated to suppliers, but simply as guidance and recommendations, thus using grammar such as “suppliers should…”.

• 3. Obligations/concrete expectations to supplier: The implementation process is formulated in an authoritative way, implying an obligation.

• 4. Obligations/concrete expectations to both supplier and company itself: The implementation process also formulates steps to be taken by the company itself, and a level of responsibility for non-compliance.

• 5. Corrective action plan: On top of the previous requirements, the implementation provision puts in place a procedure to follow in case of non-compliance. Attention: A simple termination of the contract is sufficient to be included in group 5.

For groups 3, 4 and 5, the conditions are cumulative, meaning that for instance a code’s implementation will only be rated 5 if conditions of group 4 are present. After a pilot ran by three experts on 80 companies (roughly 10% of the 810 SCCs constituting our dataset) based on an initial classification in five groups,⁴ two experts independently coded the remainder of the dataset. A total of 70 discrepancies in coding were identified between the two coders. This amounts to a 0.872 inter-reliability score according to the Cohen’s Kappa calculation, suggesting a good level of agreement between the two coders. The cases of discrepancies were discussed and resolved. Most related to some vagueness or incoherency of vocabulary used.

Results of study 1

Table 2 provides an overview of the number of management system categories included in supplier codes of conduct. As noted earlier, we distinguish between five categories of management systems to implement SCC (i.e., transparency of the supply chain, risk assessment and monitoring, training programs, corrective action processes, and reporting procedures and grievance mechanisms). Our results demonstrate that few SCCs refer to implementation mechanisms or do so superficially. Out of the 1241 companies in our sample, 810 have a SCC, of which 101 (12.5%) codes do not have any implementation provisions according to our definition. A relatively high number of 179 codes (22.1%) does not refer to any of the categories of implementation provisions. Only 46 codes (5.7%) refer to all five categories.

Table 2 Number of management system categories included in supplier codes

Figure 2 shows the percentage of codes referring to particular implementation provision categories. Risk assessment and monitoring is referred to most often (47.8% of codes), followed by corrective action plans (47.3% of codes). Few supplier codes included the importance of training in their implementation provisions (n = 189, 23.3%).

Fig. 2

Reference to management systems to implement supplier codes

A comparison of these results across different sectors and regions demonstrates that reference to specific management system categories is not affected by country or sector. Appendex 1 provides for detailed tables comparing sector and regions.

Therefore, to answer the first research question, it can be concluded that few specific management systems are included in SCCs to ensure their implementation. From our sample of 1241 companies, only 46 of them drafted a SCC that includes reference to all five relevant management system categories, namely transparency, risk assessment and monitoring, training programs on implementation, corrective action process, as well as reporting procedures and grievance mechanisms. Companies refer more often to risk assessments and monitoring or to corrective action plans than they refer to reporting procedures or trainings. Only 30% of the SCCs referred to three or more categories.

Table 3 provides an overview of the classification conducted by our expert assessment, following the analysis of the quality of SCC implementation provisions. Our findings indicate that most companies draft supplier codes of conduct to influence the behavior of their suppliers without committing themselves to taking an active role in the implementation programs. That is, 60% of companies fall in category 3, meaning that they do have implementation provisions, but those only enunciate obligations to be respected by suppliers themselves, without including a conduct to be adopted by the multinational itself. 83% of companies with a code of conduct do not include any provision that would bind themselves to implementing their programs (only reflected in categories 4 and 5).

Table 3 Quality assessment of implementation provisions in supplier codes of conduct

Table 4 highlights the geographical differences in SCC ratings. It appears that Australian companies are slightly more inclined than companies located in other regions to commit to implementing their standards, shortly followed by European companies. Lack of any implementation provisions is most common in Asia.

Table 4 Quality of implementation provisions across geographical regions

Table 5 reflects the sectorial differences of our expert ratings. Sectors particularly influenced by consumer retaliation, such as consumer discretionary and consumer staples, are more often classified as having a higher quality of codes. Appendex 1 provides for detailed tables comparing sector and regions.

Table 5 Quality of implementation provisions across sector

Our quality assessment of implementation provisions in SCCs leads us to conclude that companies are reluctant to formulate their implementation mechanisms as instruments that would bind them to act a certain way. In most cases, codes are formulated as “companies reserve the right to…”. Suppliers are at the receiving end of these codes, where clear expectations of compliance with labor standards are included, but the steps to be taken and tasks to be done by the companies themselves are limited. This tells us that SCCs are primarily a top-down policy, where companies seldomly accept to be bound by its effect. To answer the second research question, it is evident that codes’ implementation provisions put little responsibility on multinationals, especially considering that only 5% classify as what we rated the best practice scenario.

As we employed two methods to analyze the implementation provision in SCCs, we also wanted to investigate the correlation between the number of management system categories (method 1 based on text analysis) and the quality assessment of implementation provisions (method 2 based on expert assessment). We indeed found a positive and strong (r = 0.818, p < 0.001). Table 6 provides the results of the cross-tabulation of these two variables, and includes the variables related to the frequency of the keywords for each of the management system categories. This allows us to understand specifically which management system correlates the most with our assessment of SCC quality.

Table 6 Means, standard deviations and correlations between expert assessment and frequency of management system keywords

This tells us that companies referring to most implementation tools as “key words”, such as those recommended by international organizations, are most likely the same companies having a higher assessment of code implementation provisions’ quality, according to an expert classification. This correlation also confirms that two different measurement of SCCs’ implementation provisions appear to lead to similar conclusions. In other words, such high correlation between these two variables could also be interpreted as a validity test of their coding. However, this correlation does not indicate that those programs are reflected in practice, nor effectively implemented. The next section will give us more insights on the actual existence of implementation programs, when included in SCCs.

Study 2. From commitment to actions in implementation mechanisms

In this second study, we compare the content of SCC implementation provisions with the scoring of multinationals’ implementation programs by KnowtheChain. This allows us to investigate the gap between the written provisions in supplier codes of conduct and the actual existence of programs in practice. KnowtheChain is a nonprofit organization evaluating the policies and practices that companies have in place to tackle forced labor in their global supply chains. Their database consists of general and specific scores for multinationals’ global supply chain behavior, which is updated every year and publicly available on their website.⁵

Methods for study 2

KnowtheChain sample of companies

KnowtheChain includes companies based on a strict selection process, using three primary inclusion criteria: exposure to forced labor risk, market cap, and sectors under high exposure to forced labor risks. Regarding this last element, three sectors were identified, namely the information and communications technology, food and beverage, and apparel and footwear. For 2020–2021, 129 companies from all over the world were selected. From the sample of companies evaluated by KnowtheChain in year 2020–2021, we found an overlap of 94 companies with our database. From those 94 companies, only 77 had a supplier code of conduct on their website according to our database. Considering this relatively small overlap and hence small sample size for Study 2, the results presented herein should be considered preliminary insights and therefore definite conclusions are not warranted.

KnowtheChain scoring

KnowtheChain produces benchmarks to evaluate the efforts of companies in addressing forced labor in their supply chains. Based on the UN Guiding Principles on Business and Human rights, KnowtheChain evaluates corporate policy commitments, due diligence practices and remedy. Seven thematic indicators are studied: commitment and governance, traceability and risk assessment, purchasing practices, monitoring and verification, recruitment, worker voice, and remediation. The scores for each of these categories are obtained following a strict benchmark methodology, developed through consultation with stakeholders. These benchmarks are applied similarly to each company evaluated and published online.⁶ The data for the benchmarks are collected via document analysis and website scrapping for any English language disclosure by companies, additional disclosure provided by companies upon interaction, publicly available forced labor allegations from third parties sources, and companies responses to allegations and third party verification with third parties, such as the Responsible Business Alliance and the Fair Labour Association.⁷

Each company evaluated by KnowtheChain is given an overall average score from 0 to 100, as well as a score from 0 to 100 attributed to each of the above-mentioned seven categories. KnowtheChain benchmark methodologies are updated and improved every year, to fine-tune the evaluation to the latest expectations and best practices examples on companies’ policies and practices to prevent labor rights in global supply chains. For this study, we used KnowtheChain scoring and benchmark methodology from 2020–2021, considering that the SCCs in our own dataset stem from this period. Moreover, it is important to note that KnowtheChain goes beyond scoring companies simply based on the existence of CSR programs, but also assesses the quality and implementation. For instance, on the companies’ risk assessment on supply chains, KnowtheChain distinguishes between companies carrying out risk assessment (64% in 2021) and companies including workers in risk assessments (only 9%). The results of KnowtheChain analysis for 2020–2021 are detailed in the report “Closing the gap, evidence for effective human rights due diligence from five years measuring company efforts to address forced labour”.⁸ The data show slow progress in companies’ human rights due diligence efforts to tackle labor rights and may be indicative of a lack of preparedness for upcoming legislation. An interview with KnowtheChain in March 2023 allowed us to fully understand their benchmark methodology and confirm the relevance of this research with the organization.

Comparing KnowtheChain results with our database

KnowtheChain scoring provides details and information about companies’ implementation programs that complements the findings of Study 1. While our own dataset provides for textual content analysis of codes, KnowtheChain explores actual corporate actions to prevent and mitigate labor risks in global supply chains. It is relevant to compare and analyze these two sources of data, to assess whether SCC content correlates with corporate actions. We formulate the following three hypotheses:

• Hypothesis 1: KnowtheChain scores will be higher for companies which have adopted a SCC compared to companies who have not adopted a SCC.

While SCC presence has not been identified as a clear predictor of better behavior on corporate global supply chain labor performance in the literature (Vandenbroucke 2024), we expect that setting standards for labor conditions is a first step towards implementation programs and acknowledgement of responsibility for labor conditions in the global supply chain. In turn, we would expect these companies to score better in terms of efforts deployed to tackle forced labor, as assessed by KnowtheChain.

• Hypothesis 2: The quality of implementation provisions in SCCs is positively associated with KnowtheChain scores.

We expect that the quality of SCC implementation provisions, as evaluated by our experts in Study 1, is associated with the quality of actual implementation programs in place, as evaluated by KnowtheChain, because the expert assessment focused on the level of responsibility that companies commit to.

• Hypothesis 3: KnowtheChain scores will be higher for those companies who commit to establish management systems than for companies who do not.

We expect that implementation provisions will reflect implementation actions, as we accept the hypothesis that companies ‘walk the talk’. Therefore, we foresee that companies committing to set management systems to implement their codes will score higher in their efforts to tackle forced labor, as assessed by KnowtheChain.

Results of study 2

To test Hypothesis 1, we compared companies with (n = 77) and without (n = 17) a supplier code of conduct. An independent samples t-test revealed that the average KnowtheChain score between these two groups was not significantly different (p = 0.349): companies without a code reach an average global score of 32/100, while companies with a code reach an average of 34/100. Thus, although the difference is in the expected direction, there is no statistical evidence that the presence of a code indicates a better approach to preventing labor risks in global supply chains following the KnowtheChain methodology. The first hypothesis is not supported. We also did not find significant correlation with any of the subcategories rated by KnowtheChain.

The second hypothesis proposes that the quality of implementation provisions in SCCs is positively associated with KnowtheChain scores. We assessed the correlation values of our expert assessment with both the overall KnowtheChain score and the KnowtheChain scores per category, as can be seen in Table 7. None of the coefficients are significant, although positive. Given the small sample size and hence low statistical power, we find it important to highlight some of the positive correlation coefficients. For instance, the management and accountability indicator of KnowtheChain is the category that is most strongly associated with the overall quality of the SCC. The more companies commit to sharing responsibility to implement codes as per their written implementation provisions, the better their management and accountability indicators. A good management and accountability score means that companies have established clear responsibilities and accountability for the implementation of their supply chain policy.⁹ This result thus suggests that our assessment of implementation provisions matches a clearer responsibility and accountability.

Table 7 Correlation matrix between expert assessment and KnowtheChain scores

The sub-category remedy evaluated by KnowtheChain is also positively associated with our expert assessment variable. KnowtheChain rates companies higher on remedy when the company has a process to provide remedy to workers in its supply chains in case of forced labor and human trafficking. Seeing a correlation with our expert assessment confirms that companies taking a higher level of responsibility in their written statements matches a collaboration with workers, by giving them an avenue of discussion with the company itself.

Hypothesis 3 contends that reference to management systems in a SCC is positively correlated with KnowtheChain scores. We observe a positive correlation between the general KnowtheChain scores and the number of management systems, yet without statistical significance. Figures 3 and 4 visually depicts the difference of KnowtheChain scores when there is and there is no reference to each management system. The detail of all correlations, broken down per category of KnowtheChain scores, can be found in Table 8. The full correlation matrix can be found in Appendex 2.

Fig. 3

KnowtheChain global scoring when there is and when there is no reference to management systems

Fig. 4

Mean of KnowtheChain scoring in the presence or absence of keywords on Reporting. Note. Mean of KnowtheChain’s scores on each of these categories, depending on whether or not they include reporting keywords in their SCC

Table 8 Correlation matrix of the frequency of management systems and KnowtheChain scores

In relation to Hypothesis 3, it stands out that the presence of reporting procedure keywords is significantly and positively related to KnowtheChain scores: The more companies commit to setting whistleblowing and grievance mechanisms in their SCC, the better implementation efforts are identified by KnowtheChain. Allowing supply chain workers to report on labor standards issues is thus a factor of better corporate practices in their supply chain governance. Figure 4 gives a visual representation of the differences in KnowtheChain scoring between companies including reporting mechanisms in their implementation provisions and those who do not. We notice the presence of reporting keywords almost always indicate higher KtC means, in some cases more significantly so.

The frequency of keywords relating to training, meaning training of supplier regarding their minimum labor standards, is also positively associated with KnowtheChain ratings. This correlation is especially significant regarding KnowtheChain’s evaluation of supply chain traceability, i.e. the extent to which the company demonstrates an understanding of their supply chain risks and composition. Training suppliers also leads companies to be more aware of their supply chain issues. A visual representation of the mean of KnowtheChain results is given in Figs. 5 and 6.

Fig. 5

Mean of KnowtheChain scoring in the presence or absence of keywords on Training. Note. Mean of Know the Chain’s scores on each of these categories, depending on whether or not they include training keywords in their SCC

Fig. 6

Mean of KnowtheChain scoring in the presence or absence of keywords on Corrective Action Plan. Note. Mean of Know the Chain’s scores on each of these categories, depending on whether or not they include corrective action plans keywords in their SCC

In the case of the frequency of corrective action process keywords, we however notice a negative correlation with most of the KnowtheChain scores. For this method, corrective action process keywords included the termination of contract of suppliers in case of non-compliance. This means that when companies threaten their suppliers to end the relationship, KnowtheChain observes worse-off implementation efforts and practices on CSR in global supply chains. A similar negative correlation is identified regarding the frequency of risk assessments keywords: where companies often refer to supply chain audits and factory visits, KnowtheChain results are more likely to be inferior. Figure 6 gives a visual representation of the difference of KnowtheChain mean between companies including Corrective Action Plan keywords in their SCC, and those who do not. As opposed to the previous similar figures, the two lines are much more equal, and absence of the corrective action plan keywords sometimes exceeds its presence.

Therefore, our third hypothesis is partially supported. There is evidence that, when companies refer to management systems to implement their SCC, KnowtheChain scores are higher, but this relationship is not statistically significant. A significant correlation between KnowtheChain scoring and specific management systems referred in codes is identified: higher frequency of reporting procedure keywords and training keywords in SCC is correlated with better KnowtheChain scores. In contrast, higher frequency of risk assessment and corrective action plan keywords are negatively correlated with KnowtheChain scores, yet this relationship in not statistically significant.

Discussion

We began this paper by reflecting on the growing expectations towards companies to accept a responsibility to prevent and mitigate labor risks in their global supply chain, and how this led to the development of voluntary governance notably with the adoption of SCCs. The two studies conducted here contribute to our understanding of the level in which companies commit to their policy implementation.

Implementation provisions in SCC: How companies ‘talk’

While our results show that SCCs are widely adopted, they also indicate that companies only to a limited extent include implementation provisions in their supplier codes. 34.6% of codes in our sample either have no implementation provisions or have provisions that do not refer to any of the implementation mechanisms recommended by international texts. Only 5.7% of codes have implementation provisions that discuss all core mechanisms recommended by international standards – namely reference to transparency, risk assessment and monitoring, training programs, corrective action processes, and reporting procedures or grievance mechanisms.

Among management systems referred to in codes’ implementation provisions, companies most often include risk assessment and monitoring practices such as audits and factory visits, as well as corrective action plans in case of non-compliance with codes’ standards. Nearly half of the codes in our dataset include these systems. As underlined by Mamic (2005), MNEs’ auditing and monitoring programs generally involve the development of corrective action plans with supplier factories, which explains why monitoring and risk assessment often go hand in hand. However, most often codes only refer to these two implementation mechanisms – nearly 70% of codes include not more than two categories of management systems. Reference to the training of suppliers to implement labor standards, and reference to reporting procedures (e.g. whistleblowing or grievance mechanisms) are less often included in codes.

These results build on previous research (Vandenbroucke et al. 2024) that showed that nearly 90% of SCCs referred to most core ILO standards (namely prohibition of child labor, forced labor, discrimination, occupational health and safety, and to a lesser extent trade union rights). Companies are eager to set labor standards, but less so to take responsibility to implement them. This result confirms the notion of policy-practice decoupling (Bromley & Powell 2012) and suggests the existence of window dressing: while companies make promises, their actions may not reflect these commitments.

A shared responsibility and collaborative approach with suppliers: Necessary to ‘walk the talk’

Young’s model of shared responsibility calls for a collaborative approach to address supply chain issues. The premise is that actors are all connected to structural injustice and should collectively struggle against it (Young 2004). Our studies unfortunately show that it is not the approach adopted by most companies when they draft supplier codes of conduct. By analyzing the language adopted in codes of conduct, we observe that SCCs most often articulate expectations for suppliers, but rarely phrase implementation provisions as a responsibility pending (also) on themselves. Only 17% of SCCs appear to bind the multinationals to actions of implementation, reflecting a shared responsibility.¹⁰

Moreover, we observe that SCCs more often include references to management systems favoring the ‘compliance’ approach and are less incline to adopt the peer-to-peer governance (Jiang 2009) favoring stakeholder involvement and workers’ voices. For instance, audits and supplier monitoring are the most common management system referred to, while they are criticized in the literature for reproducing dynamics of dominance and hierarchy between the buyer and the supplier. Corrective action plans, where buyers have the leverage to “drop” the supplier in case of noncompliance, provoke a similar effect – and yet are the second most common instrument mentioned in SCC. This develops a top-down governance by multinationals (Lindholm et al. 2016). In fact, our results show that referring to suppliers’ surveillance via audits, or the termination of contract, seem to deteriorate implementation programs in practice, as rated by KnowtheChain. In contrast, grievance mechanisms are less referred to in codes – while they have proven effective in ensuring a discussion at different organizational levels, if workers are protected when speaking out (Singh 2011). When SCC refer to grievance mechanisms, such as whistleblowing procedures and hotlines, KnowtheChain scores are also higher, proving better efforts and programs developed in practice.

We also find no evidence in support of that the claim that the adoption of a SCC—whether it includes labor standards and implementation mechanisms or not—ensures good corporate social behavior in global supply chains. This result corroborates previous empirical findings (Distelhorst et al. 2015; Barrientos & Smith 2007), this time using KnowtheChain data. Companies with a code do not score significantly better on preventing and mitigating forced labor risks in their supply chains, compared to companies without a code.

Therefore, while our findings should be interpreted carefully in light of our methods’ limitation, they contribute to a small but growing empirical literature and confirm the necessity for companies to adopt the collaboration approach with their suppliers. To prevent policy decoupling and to ‘walk the talk’, it is not sufficient to adopt a code and commit to implementing it. Code’s implementation should integrate workers’ voices, ensure supplier engagement, and accept a share of responsibility in its non-compliance. A top-down approach of suppliers’ surveillance has not proven fruitful in preventing forced labor down the supply chain.

Limitations and opportunities for future research

These studies are pioneering in the combination of the methods and databases, and aim at answering a recurring question in CSR: Do companies walk the talk, accept a level of responsibility, and act upon their words? Since this debate requires data for better regulation of corporate responsibility in global supply chains, our empirical studies make a timely contribution to this field. However, our results should be interpreted in light of their limitations, both in terms of the sample studied and the methods employed.

Firstly, SCCs are only one of the CSR policies developed by multinationals. Other documents, such as purchasing contracts with suppliers, ethical charters, press releases, and meeting notes, may reflect corporate efforts to implement labor standards. The object of our study limits corporate “talk” to their supplier codes provisions, which arguably are not the most reliable source of information on corporate practice. In the future, a thorough study of companies’ commitments in and outside of their SCCs would provide a more comprehensive approach of investigating companies’ private governance on labor issues.

Secondly, the expert assessment, although cross-checked with different experts to ensure reliability, is a subjective method for classifying codes of conduct that we developed ourselves. Although allowing us to assess a specific variable – e.g. corporate responsibility commitment—this limits the reproducibility of this study. Future research could base their quality assessment on existing benchmarks, or in collaboration with international institutions likely to reproduce the study over time.

Thirdly, an important limitation of the results of Study 2 is related to the small sample size. Not only did our correlational analyses suffer from low statistical power, but we also do not know how representative the companies selected by KnowTheChain for that year were for our bigger dataset (of 1241 companies). Moreover, the bivariate character of the analysis does not warrant causal claims and does not exclude the possibility that other factors have a confounding effect. That being said, given that KnowtheChain benchmarks contain a wide array of information and variables, we encourage future research in this field to draw on these data to gain understanding of companies’ efforts to tackle forced labor in global supply chains.

Finally, it is important to note that SCC content analysis may not reflect the reality of implementation programs in place, but only give an indication on how companies communicate with their external stakeholders on their implementation provisions. To fill this gap, an interesting next step to this study would be to conduct qualitative research on what business practices are effectively in place on the basis of companies’ codes. Here, it is relevant to interview supply chain human rights advisor of multinationals, to gain in-depth understanding on the translation from policy to practice and its challenges.

Practical implications

Our results offer insights to policy makers regulating corporations in their global supply chain approach, as well as to corporations. They suggest that, while companies set labor standards, their implementation is sub-optimal and companies do not often ‘walk the talk’ with implementation programs likely to effectively improve workers’ conditions. This proves the importance of developing a legal framework binding companies to act upon their SCC, and invest in the implementation of their labor standards down the supply chain. Currently, a legal framework on mandatory human rights due diligence is developing, which a hopeful step towards a better share of responsibility among the supply chain actors to protect workers’ rights. Based on our results, to make a difference, policy makers should guide multinationals towards a collaborative approach with their suppliers, notably by promoting management systems ensuring stakeholder engagement and drifting away from a top-down approach of corporate self-regulation. On their end, to apply actions to their commitments, companies should accept a level of responsibility, including in the content of their policies. To effectively ‘walk the talk’, companies should not only surveille and monitor their suppliers, but more importantly invest costs, discussions, and training with their suppliers. We recommend that, instead of supplier audits, companies favor whistleblowing hotlines and trainings and that, instead of relying on the threat of corrective action plans ultimately terminating supplier contracts, they engage with their suppliers and participate in the costs of implementation.

It remains to be seen whether the due diligence legislations developing, such as the European corporate sustainability due diligence Directive, will promote the shared responsibility and collaborative approach, to address the pitfalls of the current corporate practices and bridge the gaps between the standards setting and the actions.

Appendices

Appendix 1

Results of study 1 broken down per sector and geographic location

Table 9 Crosstabulation expert assessment with geographic location

Table 10 Crosstabulation expert assessment with sectors

Table 11 Crosstabulation number of management systems with geographic location

Table 12 Crosstabulation number of management systems with sectors

Appendix 2

Full correlation matrix for study 2

	KTC Global score	KTC Management and accountability	KTC Training	KTC Stakeholder engagement	KTC Traceability	KTC Risk Assessment	KTC Purchaasing practices	KTC Worker Engagement	KTC Freedom of association	KTC Grievance mechanisms
KTC Global score	1
KTC Management and accountability	0.690**	1
KTC Training	0.841**	0.670**	1
KTC Stakeholder engagement	0.816**	0.557**	0.734**	1
KTC Traceability	0.673**	0.389**	0.459**	0.559**	1
KTC Risk assessment	0.820**	0.626**	0.681**	0.725**	0.514**	1
KTC Purchasing practices	0.745**	0.536**	0.581**	0.673**	0.511**	0.587**	1
KTC Worker Engagement	0.764**	0.379**	0.593**	0.643**	0.475**	0.622**	0.580**	1
KTC Freedom of association	0.545**	0.189	0.311**	0.382**	0.222*	0.445**	0.606**	0.616**	1
KTC Grievance mechanisms	0.786**	0.564**	0.628**	0.672**	0.465**	0.623**	0.605**	0.617**	0.485**	1
KTC Monitoring	0.914**	0.602**	0.786**	0.673**	0.542**	0.710**	0.604**	0.696**	0.454**	0.659**
KTC Corrective action plans	0.842**	0.684**	0.734**	0.624**	0.510**	0.698**	0.559**	0.556**	0.342**	0.604**
KTC Remedy	0.883**	0.691**	0.757**	0.678**	0.568**	0.710**	0.594**	0.556**	0.348**	0.664**
Implementation keywords: Number of catgories referred to	0.119	0.149	0.033	0.155	0.189	0.107	0.111	0.081	-0.087	0.092
Frequency of ‘Transparency’ keywords	0.098	-0.029	-0.010	0.101	0.241*	-0.058	0.090	0.068	-0.100	-0.010
Frequency of 'Risk assessment' keywords	0.022	0.039	-0.002	0.150	-0.082	0.023	-0.033	0.003	-0.079	0.158
Frequency of 'Training' keywords	0.214	0.108	0.012	0.166	0.306**	0.206	0.054	0.165	-0.092	0.265*
Frequency of 'Corrective action process' keywords	-0.052	0.060	-0.047	0.063	-0.013	-0.040	-0.111	-0.049	-0.175	0.063
Frequency of 'Reporting procedures' keywords	0.245*	0.203	0.184	0.117	0.237*	0.209	-0.018	0.187	0.057	0.174
Expert assessment	0.109	0.195	0.047	0.134	0.075	0.173	0.136	0.099	0.037	0.121

	KTC Monitoring	KTC Corrective action plans	KTC Remedy	Implementation keywords: Number of categories present	Frequency of ‘Transparency’ keywords	Frequency of 'Risk assessment' keywords	Frequency of 'Training' keywords	Frequency of 'Corrective action process' keywords	Frequency of 'Reporting procedures' keywords	Expert assessment
KTC Global score
KTC Management and accountability
KTC Training
KTC Stakeholder engagement
KTC Traceability
KTC Risk assessment
KTC Purchasing practices
KTC Worker Engagement
KTC Freedom of association
KTC Grievance mechanisms
KTC Monitoring	1
KTC Corrective action plans	0.809**	1
KTC Remedy	0.820**	0.935**	1
Implementation keywords: Number of catgories referred to	0.043	0.110	0.137	1
Frequency of ‘Transparency’ keywords	0.097	0.085	0.101	0.579**	1
Frequency of 'Risk assessment' keywords	-0.042	0.103	0.075	0.283**	0.150**	1
Frequency of 'Training' keywords	0.164	0.144	0.134	0.498**	0.366**	0.420**	1
Frequency of 'Corrective action process' keywords	-0.147	0.025	0.055	0.470**	0.193**	0.570**	0.252**	1
Frequency of 'Reporting procedures' keywords	0.223	0.200	0.170	0.386**	0.066	0.016	0.077*	0.075*	1
Expert assessment	0.042	0.154	0.176	0.818**	0.216**	0.276**	0.193**	0.329**	0.185**	1

1. * indicates statistical significance at the p < 0.05 level
2. ** indicates statistical significance at the p < 0.01 level