Detecting and remediating modern slavery in supply chains: a targeted audit approach

Abstract This paper investigates modern slavery detection and remediation. Action research has been conducted in the textiles and fashion industry, with the primary engagement involving a multi-billion pound (GBP) turnover company and their modern slavery investigation at a high-risk supplier in South East Asia. This paper responds to calls from the literature to investigate the modern slavery detection process and provides empirical evidence involving collaboration with a large multinational NGO and another of the audited supplier’s customers. Findings are presented from a first-hand account of the detection process and suggest that a targeted audit is more likely to identify key indicators of modern slavery. This type of audit includes investigating the end-to-end recruitment process by using a parallel structure of management and worker interviews and documentation review. Evidence is also provided of the company’s remediation process, which includes partnering with a local NGO to empower workers and collaboratively develop suppliers.


Introduction
Modern slavery, as occurring in the context of supply chains, has been defined as: 'the exploitation of a person who is deprived of individual liberty anywhere along the supply chain from raw material extraction to the final customer for service provision or production" (Gold, Trautrims, and Trodd 2015, 487). The extent of this contemporary phenomenon is emphasised by recent research conducted by the International Labour Organisation (ILO) and the Walk Free Foundation that estimated that, globally, 16 million people are victims of economic forced labour exploitation, including through modern day slavery in supply chains (ILO (International Labour Organisation) 2017). In particular, according to the recent 'Global Slavery Index 2018' published in July 2018, the textiles and fashion industry is one of the largest perpetrators of modern slavery globally (Walk Free Foundation 2018). Modern slavery is thus a global issue with instances in both developing and developed countries. For example, there have been recent cases reported in the UK news such as of 'a slave workforce' being discovered at Kozee Sleep, a UK bed manufacturer supplying well-known high street retailers (BBC 2016) and human trafficking discovered in a Sports Direct warehouse (The Guardian 2017; BBC 2019).
Legislation has begun to be introduced around the world in response to the modern slavery threat. Recent examples include the 'California Transparency in Supply Chains Act' (State of California Department of Justice 2010) and the 'UK Modern Slavery Act' (UK Government 2015). The UK legislation contains the 'Transparency in Supply Chains' clause that requires organisations with a turnover in excess of the currently stated threshold of £36 million to publish a statement each financial year regarding action that is being taken to combat modern slavery in their supply chains. The introduction of modern slavery legislation has provoked discussion of this issue in the broader management literature (e.g. Crane 2013), and two recent key conceptual papers by New (2015) and Gold, Trautrims, and Trodd (2015) along with empirical research by Benstead, Hendry, and Stevenson (2018) have contributed to a growing literature on sustainable supply chain management (SSCM). It has been acknowledged that slavery can enter the supply chain where labour intensity is high and profit margins low (Crane 2013). A growing body of research has also considered the heightened modern slavery risk when production and labour recruitment is outsourced (Allain et al. 2013;LeBaron and Lister 2015). Amongst others, The Ethical Trade Initiative (ETI) has provided guidance for companies to improve their awareness of deception and coercion during the sub-tier recruitment process. Practices identified that lead to forced labour is, for example, withholding wages, recruitment fees, and passport retention (ETI 2017). According to the ILO (International Labour Organisation) (2019), 'The retention of passports and other identity documents is one of the most common forms of coercion, restricting a migrant worker's freedom of movement, preventing them from seeking help and trapping them in forced labour'.
Amongst the organisational responses to this new legislation, as disclosed in modern slavery statements, is the use of supplier audits to detect modern slavery (Stevenson and Cole 2018). Whilst it is well documented that auditing is commonly used to investigate social standards in a factory (Jiang 2009;Lund-Thomsen and Lindgreen 2014;Helin and Babri 2015), many authors have questioned the effectiveness of audits (Prieto-Carr on et al. 2006;Tallontire et al. 2011;Gu enin-Paracini, Malsch, and Paill e 2014;Power and Gendron 2015). In particular, both New (2015) and Gold, Trautrims, and Trodd (2015) highlighted that the illegal and hidden nature of modern slavery makes detection difficult in fragmented multi-tier supply chains. These specific issues deserve attention, yet research into modern slavery from a supply chain perspective is limited, with authors calling for more empirical work to identify effective means of detection and remediation (e.g. Gold, Trautrims, and Trodd 2015). Modern slavery is an important phenomenon to investigate in order to advance the field of Operations and Supply Chain Management as it relates to a number of important streams of literature in the field. It can be seen as a particular branch of the literature on socially Sustainable Supply Chain Management (SSCM) and a threat that represents a specific type of supply risk. Modern slavery legislation, such as that introduced in the UK, questions the sourcing practices of buyer firms and seeks to incentivise focal firms to improve the transparency of their supply chains while remediation speaks to the literature on supplier development practices. Given that organisations claim to be using audits to address socially SSCM related modern slavery risks (Stevenson and Cole 2018), it is argued here that one such line of muchneeded research is to determine whether a new form of audit process could be used that is better able to detect modern slavery, alleviating at least some of the pitfalls associated with the approach. We therefore seek to answer the following research question: How can audits be improved to better detect modern slavery in the supply chain, and how can an appropriate remediation plan be established when modern slavery is discovered?
To answer the above research question, we present empirical evidence from an action research project in the textiles and fashion industry with a multi-billion pound (GBP) turnover company, hereafter referred to as Buyer A. The first implementation of an audit process targeted at modern slavery detection at a high-risk factory in South East Asia (Supplier) is investigated, beginning with the initial preparation for the audit through to the post-audit follow-up. The first author was actively engaged in this process, which involved Buyer A collaborating with a large multinational NGO (NGO A) and another of Supplier's key customers (Buyer B). Together, Buyer A and Buyer B represent 80% of Supplier's business and Buyer B was invited to observe the audit process and add leverage. Empirical evidence is then used to understand Buyer A's ongoing remediation involving a local NGO (NGO B).
The remainder of this paper is organised as follows. First, a review of the relevant literature is provided in Section 2. The research method is then outlined in Section 3 followed by the findings in Section 4, which are then discussed in Section 5. The paper is drawn together in a conclusion in Section 6, including implications for practice and future research.

Literature review
The literature review is divided into three sections. The first section looks at transparency in supply chains, given that supplier audits are a means of gaining transparency. The second section focuses on detection using supplier auditing, i.e. one of the main tools employed in practice for achieving supply chain transparency (Egels-Zand en, Hulth en, and Wulff 2015), by reviewing the prior literature to understand the suitability of auditing for the context of modern slavery and identify the pitfalls associated with the approach. The third section focusses on remediation.

Transparency in supply chains
Although there is extant literature concerning supply chain transparency in terms of information disclosure between a buyer and supplier, there is limited literature on the public disclosure of supply chain information (Doorey 2011;Mol 2015). Some of these studies have considered voluntary disclosure when there is no regulation but stakeholder pressure (e.g. Doorey 2011; Kozlowski, Searcy, and Bardecki 2015;Gualandris et al. 2015) and evaluated its impact on competitiveness (Chen and Slotnick 2015). For example, Doorey (2011) considered the steps that led companies such as Nike and Levi Strauss & Co. to voluntarily expose their supplier lists (e.g. NGO pressure) and although the research did not analyse whether this has led to improved conditions in factories, it is suggested that this level of transparency should encourage better management of labour practices.
Given the new modern slavery legislation, there has been an increase in supply chain transparency through the disclosure of information within the annual statements published, as analysed for 101 organisations in the textiles and fashion industry by Stevenson and Cole (2018). This work identified how organisations are detecting and remediating modern slavery in terms of what they have decided to publically disclose. It provides a breadth of understanding but it is unable to go into any depth on a particular organisation. Further, it is only able to report on what organisations chose to disclose about what they are doing. It is concluded that further in-depth research, such as using engaged research approaches, is needed to study first-hand some of the practices that companies are claiming to use to increase transparency.
Appropriate actions to improve transparency have been discussed by many authors, including Macchion et al. (2018) and Cole and Aitken (2019). The former include supplier audits as a means of gaining transparency. They suggest that companies with the most advanced strategies for achieving sustainability in fashion supply chains use audits to verify the supplier's respect for environmental and social goals. In contrast, the latter argue that transparency is better gained by suppliers being more forthcoming and willing to share objectives, thereby striving for mutual goals, and that this would reduce the need for audits. However, Cole and Aitken (2019) fall short of arguing that the need for audits will be eliminated altogether, and thus as both of these recent empirical papers suggest that audits are an ongoing means of gaining transparency, it is argued here that appropriate use of audits is in need of further research particularly in the context of modern slavery.

Supply chain auditing background
There is a substantial body of literature on global supply chains, value chains, and networks in relation to labour exploitation (LeBaron, Lister, and Dauvergne 2017;Crane et al. 2019). This has, for example, discussed policies and legislation, supply chain dynamics and business models. This research identifies a number of current practices that seek to improve labour standards throughout the supply chain. For example, Barrientos (2008) reported that retailers and buyers have introduced codes of conduct for suppliers as a result of NGO pressure to improve labour standards when state regulation is insufficient to protect workers and law enforcement is weak. These codes of conduct are commonly used to manage, monitor, and control suppliers through auditing against the code (Helin and Babri 2015;Jiang 2009;Pedersen and Andersen 2006). They are also referred to as the 'compliancebased paradigm' (Lund-Thomsen and Lindgreen 2014). Advocates such as NGOs have legitimised audits as necessary forms of governance encouraging improved standards whilst sustaining good buyer/supplier reputations (LeBaron, Lister, and Dauvergne 2017).
A standard social audit applies public or private standards by investigating working conditions within a factory by touring the site, reviewing documents, and interviewing workers to further understand employee wages and benefits, working hours, harassment, health and safety, and the use of child labour (Huq, Stevenson, and Zorzini 2014;Helin and Babri 2015). These can be announced or unannounced to uncover the real conditions in a factory and conducted by brands internally, NGOs or third-party auditors (O'Rourke, 2003;Locke, Qin, and Brause 2007;Huq, Stevenson, and Zorzini 2014;Winter and Lasch 2016) yet they have a number of shortcomings, as discussed below.

Shortcomings of supply chain auditing
Although supplier audits and certifications are common practice for monitoring supply chains, there is increasing scepticism around their effectiveness (LeBaron, Lister, and Dauvergne 2017). On the one hand, the extant literature provides evidence that they can have a positive impact by adhering to or exceeding the minimum wage, improving health and safety, reducing overtime (Barrientos and Smith 2007), and improving productivity (Huq, Stevenson, and Zorzini 2014). However, research also suggests that audits that take place to ensure codes of conduct are adhered to are not always sufficient for identifying issues and improving working conditions (Huq, Stevenson, and Zorzini 2014;Lund-Thomsen and Lindgreen 2014;Egels-Zand en, Hulth en, and Wulff 2015).
Multiple studies have recognised the implications of a 'top down approach' whereby brands impose western standards on suppliers in developing countries operating in different cultural and socio-economic conditions by applying their standardised codes in various different countries (Egels-Zand en, Hulth en, and Wulff 2015; Helin and Babri 2015;Huq and Stevenson 2020). Further, this encourages a 'tick box' process which has been proven to misrepresent the conditions of a factory resulting in 'passed' audits that have later been found to have serious violations (LeBaron and Lister 2015; Outhwaite and Martin-Ortega 2019). There is also reference to 'audit fatigue' (Marshall et al. 2016) due to the constant auditing that factories face from their multiple customers. Additionally, suppliers are often responsible for the payment of audits (and follow-up audits to check noncompliances have been corrected) whilst being under the threat of order volume being withdrawn for non-compliant behaviour (Jiang 2009). Furthermore, the lack of consistency amongst codes of conduct can result in 'compliance limbo' due to conflicting brand requirementsa commonly used example of this is the different height expectations for fire extinguishers (Locke, Qin, and Brause 2007). Buyers' poor communication with suppliers limits their ability to fully understand such challenges resulting in suppliers playing a passive role. Additionally, suppliers are often overwhelmed, receiving limited support to fully understand and meet codes of conduct and the subsequent lengthy audit corrective action plans (Gould 2005;Jiang 2009).
Research has considered how current auditing strategies lead to mock compliance. Jiang (2009) for example developed and tested a conceptual model that uses transaction cost economics to explain the risk of market governance that encourages a culture of 'passing the audit' through dishonesty rather than improving standards. Suppliers commit audit fraud by hiding information through the falsification of documents, keeping separate records for auditors (double booking), and coaching workers for interviews (Egels-Zand en 2007;Jiang 2009;Huq, Stevenson, and Zorzini 2014;Plambeck and Taylor 2016;Huq and Stevenson 2020). For example, in their research on socially sustainable practices in the garment industry in Bangladesh, Huq and Stevenson (2020, p437) provided empirical evidence of mock compliance which has been categorised into 'hiding violations, short-term superficial conformance, and blatant cheating'. However, they also found that this was sometimes overlooked by buyers due to the difficulty of eradicating some issues. Instead, a developmental approach is taken to avoid further deceit. Furthermore, in their previous research Huq, Stevenson, and Zorzini (2014) found that suppliers in Bangladesh claimed buyers were only interested in improving their reputation and avoiding bad publicity rather than being serious about addressing social sustainability concerns. Similarly, authors have considered the limitations due to conflict of interest. For example, the risk of in-house auditors not disclosing violations and the limited reliability of thirdparty auditors given that they may rely on continuing to work with the lead firm (O'Rourke, 2003;Locke, Qin, and Brause 2007).
The tone of the above discussion on mock compliance could be argued to imply that the blame for ongoing modern slavery concerns in the supply chain lies entirely with the suppliers. To counter this implication, it is important to also draw on the extant literature that addresses issues of power in the supply chain in the context of an audit. This literature indicates that the textiles and fashion industry has experienced extensive offshoring as a result of buyers in developed countries (the global north) sourcing from developing countries (the global south) in search of lower costs (Christopher, Lowson, and Peck 2004;Barrientos, Kothari, and Phillips 2013). This has led to Northern powerful buyers introducing codes of conduct to address labour conditions in the global south (Lund-Thomsen and Lindgreen 2014; Ozkazanc-Pan 2019). Drebes (2016) refers to this as the 'powerful and the powerless', and it has been argued that such imbalances in power can have an impact on raising labour standards and lead to forced labour (Crane et al. 2019). Through their purchasing practices, buyers exert pressure on their suppliers to both meet their commercial and social standards (Gereffi 1999;Banerjee 2014). Amongst others, Barrientos (2008) has therefore argued that this imbalance of power needs to be acknowledged when considering corporate social responsibility practices. The power asymmetry can be seen in the 'paternalistic perspective' of codes of conduct implemented by buyers and their understanding of what they believe to be 'in the best interests of the workers' (Barrientos and Smith 2007, 725). Furthermore, it is widely acknowledged that different buyers have their own audit procedures and the less powerful suppliers are often not included in the planning of how an audit should be conducted or unable to provide input into determining what measures are achievable (Huq, Stevenson, and Zorzini 2014;LeBaron, Lister, and Dauvergne 2017). Gould (2005) thus refers to suppliers as 'passive players' in the auditing process. The disproportional power balance therefore affects the workers, and this compliance approach does not empower them or help them to have their voice heard (Barrientos and Smith 2007).
From a modern slavery perspective, there is a growing body of evidence suggesting that auditing is ineffective. In line with the broader discussion on power imbalances summarised above, there is, for example, concern that such audits are biased, benefiting the powerful organisation rather than sufficiently addressing labour exploitation within complex global supply chains (LeBaron, Lister, and Dauvergne 2017). Research has highlighted that audits are limited in scope and do not highlight cases of modern slavery, especially in the case of the more vulnerable casual and migrant workers (Barrientos, Kothari, and Phillips 2013;New 2015). These workers are often recruited through exploitative thirdparty recruitment agencies (Barrientos, Kothari, and Phillips 2013), and these agents are outside the scope of standard supplier audits (LeBaron and Lister 2015). Audits measure what is happening at one moment in time within the factory and further research is needed to consider the wider context of how a worker has entered the factory, to identify potential modern slavery risks. Modern slavery therefore expands the scope of a standard 'one-size-fits-all' audit procedure (Barrientos, Kothari, and Phillips 2013;Gold, Trautrims, and Trodd 2015;New 2015) and there is a clear need to research whether a new audit process can be used to detect modern slavery whilst taking into account the issues outlined above.

Improving the auditing process
The extant literature argues that the current auditing system needs improving by shifting from 'arm's length' auditing through market governance to supplier development (Gould 2005;Pedersen and Andersen 2006;Lund-Thomsen and Lindgreen 2014). Increased involvement from buyers has been encouraged, improving a supplier's ability to understand codes of conduct and implement any required corrective action. In their study, Huq, Stevenson, and Zorzini (2014) found that supplier development was well received by suppliers and concluded that a combination of monitoring, trust, and development were required. Similarly, Jiang (2009) and Egels-Zand en, Hulth en, and Wulff (2015) claimed that a combination of monitoring and cooperation is needed. Jiang (2009) argued that long-term contracts alone will not lead to supplier commitment to codes of conduct and instead a partnership approach is encouraged whereby suppliers play a more active role in establishing achievable targets and are provided with assistance in meeting them. Lund-Thomsen and Lindgreen (2014) referred to the 'co-operation model' that encourages a developmental approach amongst multistakeholder networks, e.g. NGOs, working with buyers and suppliers to improve social standards through monitoring and training. There is also evidence of buyers partnering with NGOs to oversee audits. Plambeck and Taylor (2006) argued that the supplier is less likely to commit audit fraud in the presence of an NGO. Similarly, in their conceptual model of modern slavery supply chain challenges, Gold, Trautrims, and Trodd (2015) drew attention to supplier development and also highlighted the need for researchers to investigate the benefit of collaboration, e.g. with NGOs, for the detection of modern slavery. The literature for example acknowledges the mediating effect that NGOs can have to help move beyond a top down approach (Barrientos and Smith 2007). Such research should however also consider the impact of the power of third party collaborators (Banerjee 2014). For example, LeBaron, Lister, and Dauvergne (2017) highlighted the increased power of NGOs as a result of the 'audit regime' as well as the profit that they can make.
Overall, there is agreement that workers' voices must also be heard (Prieto-Carr on et al. 2006). Outhwaite and Martin-Ortega (2019) refer to this as 'worker driven monitoring' and encourage the involvement of workers throughout the process from the design of improvement initiatives through to remediation. For example, they consider the importance of empowering workers through training enabling them to recognise and identify violations. Further investigation is however needed to understand how this model can be developed in practice. Therefore, given the pitfalls of auditing, there is scope to research how the process can be improved and whether alternative approaches are suitable for detecting modern slavery.

Remediation
Company responses to the detection of slave labour have not been addressed at length within the extant literaturean exception being the identification of remediation practices in a review of recently published modern slavery statements in the UK (Stevenson and Cole 2018). Action plans and follow-up audits have for example taken place for non-compliant suppliers. Additionally, organisations have engaged in firm level and supplier level development such as remediation training and collaboration with other buyers and NGOs. But further research is needed that examines firsthand how organisations are approaching remediation when modern slavery is detected.
More broadly in the SSCM literature, authors have identified actions in response to social issues in supply chains such as the termination of business, increased auditing to check on improvements, and supplier development, e.g. training (Jiang 2009;Gimenez and Tachizawa 2012;Blome, Hollos, and Paulraj 2014;Sancha et al. 2015;Zhang, Pawar, and Bhardwaj 2017). There is also the suggestion that auditing could be eliminated through stronger collaboration with suppliers (Awaysheh and Klassen 2010). Additionally, Pagell and Wu (2009) have considered collaboration with NGOs, regulators, competitors and members of the community. There is however limited research on collaboration with NGOs from a social sustainability perspective (Zorzini et al. 2015) with such research focussing instead on environmental sustainability (e.g. Albino, Dangelico, and Pontrandolfo 2012;McDonald and Young 2012). An exception is in the context of the Bangladesh garment industry. Following the Rana Plaza disaster, Huq, Chowdhury, and Klassen (2016) found that buyers had engaged with NGOs in joint initiatives to improve industry social standards, e.g. training and skills development for workers. Additionally, buyer-consortium audits have been developed involving multiple buyers, and both global and local unions supported by NGOs (Huq, Chowdhury, and Klassen 2016;Huq and Stevenson 2020). These have extended buyer audit capabilities by sharing capabilities to audit fire, electrical and structural safety standards to improve a supplier's social performance.
In the context of modern slavery, Gold, Trautrims, and Trodd (2015) drew attention to supplier development and highlighted the opportunities to study collaboration, e.g. with NGOs, for the remediation of modern slavery. Meanwhile, New (2015) argued that modern slavery remediation is distinctly different, highlighting that the illegality of modern slavery means that the conventional top down 'improvement notice' response to audit non-compliances are impractical, instead requiring the involvement of authorities and the termination of business. Yet, Stevenson and Cole (2018) found there were few instances of organisations reporting violations to authorities in their modern slavery statements. Gold, Trautrims, and Trodd (2015) also considered the detrimental socio-economic effects that could result from withdrawal from a region or country and, as a result, recommended supplier development. There are therefore contrasting views in terms of how firms should remediate modern slavery. Further, although remediation practices have been identified from publicly disclosed statements (Stevenson and Cole 2018), this is limited to the evidence that organisations have chosen to share. There is therefore scope to develop a deeper and richer understanding of the remediation process that is taking place following the identification of modern slavery.
Research has also considered the role of the buying firm and their purchasing practices that can inadvertently lead to exploitation (Barrientos 2008). Anner (2019) for example investigated this in the Bangladesh garment export sector and considered the demand for low prices, lead time pressure and order volatility which results in 'squeezing workers' rights'. New (2015) argued that modern slavery is 'generated by the normal system' by referring to the 'right hand' establishing corporate social responsibility policies whilst the 'left hand' can cause modern slavery due to commercial purchasing practices. Stevenson and Cole (2018) also identified organisations disclosing that they had introduced purchasing practices training as part of their modern slavery remediation and suggested that modern slavery legislation could encourage awareness beyond the procurement function to the boardroom level. This supports findings in the SSCM literature that consider the need to embed sustainability into organisations with support from top management (Pagell and Wu 2009;Pagell and Shevchenko 2014). Thus, modern slavery remediation requires investigation as part of due diligence within SSCM and research needs to understand how this is addressed given the commercial power of retailers and brands (Gereffi 1999).
There are therefore two important research gaps: 1. To investigate audit methods and explore whether better approaches can be found for the detection of modern slavery. 2. To understand the strategies that firms employ when modern slavery is detected.
In addressing these research gaps, it will be important to consider how to balance the concepts of monitoring, trust and development (Huq, Stevenson, and Zorzini 2014); whether NGOs can be effectively engaged in deploying these concepts, and how the role of power and control plays out amongst the various stakeholders (LeBaron, Lister, and Dauvergne 2017); and whether it is possible to more readily hear the workers' voice (Prieto-Carr on et al. 2006) in the context of modern slavery detection and remediation.

Action research approach
This study has used action research, a qualitative approach that has become increasingly prevalent in the study of organisations. It involves the engagement of the researcher and assumes social phenomena are continuously changing (Coughlan and Coghlan 2016;Easterby-Smith et al. 2018). Coughlan and Coghlan (2016) describe action researchers as 'outside agents who act as facilitators of the action and reflection within an organisation'. Therefore, in contrast to other approaches, the researcher does not keep a distance from the subject being researched and often participates in the change process, thus enabling deep learning about the organisation (McKay and Marshall 2001;Easterby-Smith et al. 2018). The researcher is therefore simultaneously taking action and creating knowledge (Coughlan and Coghlan 2016).
In the context of SSCM, there have been calls for researchers to use innovative engaged methodologies such as action research (Touboulic and Walker 2016). Arguably this is particularly important when investigating a complex social issue such as modern slavery. Action research is fundamentally about change, with this research focussed on changing the audit process. There were two objectives, the action itself in terms of the audit process and reflection on this as it happened (Coughlan and Coghlan 2016).
Adopting this research approach has enabled the gathering of rich data and access to privileged information, including Buyer A's detailed plans regarding modern slavery detection and remediation involving collaboration with Buyer B, NGO A and NGO B. This was facilitated by the first author's previous relevant international industry experience, which enabled trust to be built quickly. It also meant that there was the necessary level of expertise to enable the gathering of detailed information from the outset. The researcher was also mindful of power imbalances and has had extensive experience building relationships with international suppliers and interacting with factory workers (Sultana 2007;€ Ozkazanç-Pan 2008). It is important to acknowledge the positionality of the researcher as a white western female researcher, yet the first author's previous international industry experience and time spent living in a less developed country, has shaped her world view. This positionality has therefore influenced the fieldwork and the development of the paper providing insights into the perspectives of local people living and working in the global south and experience in gaining the trust of factory workers in this context (England 1994). It is argued that action research has resulted in access to sensitive information that would not have been achieved through an alternative method such as case study research. This is due to the ability to participate in the action taken to tackle modern slavery within the focal company, Buyer A.
The level of trust and strength of the relationship resulted in the first author participating in the targeted modern slavery audit in South East Asia. Action research enabled the researcher to be fully engaged in the trip, which also involved interviewing the migrant workers. This meant the researcher had the same level of involvement as Buyer A's employees, which then led to in-depth discussions due to a mutual understanding of the experience. As a result, the researcher learnt from first-hand conversations with the supplier and migrant workers rather than hearing secondhand accounts from Buyer A. This helped to mitigate any power imbalances in the interpretation of information. The researcher was actively involved in the audit planning, the audit itself (including all team briefing, interview preparation and debriefing sessions), and many of the post-audit remediation activities. For example, the audit protocol was cocreated (see Appendix) as was the 'Modern Slavery Toolkit' produced at the remediation stage. Additionally, the researcher was involved in shaping the public reporting of the audit in Buyer A's modern slavery statement. There was therefore an awareness of the power of the researcher throughout the process and, for example, their involvement in allowing the workers to express their own views (Sultana 2007).
Action research is an emergent process (Coughlan and Coghlan 2016). Although the project was outlined at the beginning, it was uncertain how Buyer A's approach to modern slavery would unfold. Thus the action was motivated and determined by the research, and Buyer A actively sought the collaboration of research partners given the need to develop new ways of working. It was therefore important for the researcher to gather all information and be adaptable as the project progressed. The targeted audit emerged as one of five initiatives developed as part of the research and the chosen supplier was identified through a risk matrix cocreated by the researcher.
To prevent researcher bias, regular fortnightly meetings took place with all three researchers and quarterly meetings took place between the researchers and a key representative from Buyer A. Thus, whilst one researcher was heavily involved with Buyer A, the other two had less contact enabling them to look out for signs of the first researcher 'going native' and to counteract any such issues that arose. Quarterly reports were also prepared, allowing the researchers to feedback on key findings. This helped to ensure there was cyclical analysis between data and theory. Additionally, the work was presented to external audiences at the university and international conferences to step back from the field and discuss the work with those not involved in the research.

The focal company: Buyer A
The researchers have engaged with the evolving modern slavery detection and remediation practices of Buyer A, a UK based, multi-billion pound turnover company in the textiles and fashion industry. This company was chosen as it is in an industry characterised by complex, global supply chains and high labour intensity, which makes it vulnerable to modern slavery. The nature of the relationship was such that, in effect, one of the researchers worked part-time for Buyer A over a 20-month period (on an expenses only basis, without being employed).

Data collection
Much of the data collection surrounds the first implementation of Buyer A's newly developed audit process targeted at modern slavery detection; inclusive of the initial preparation, through to the follow up. This targeted audit was both a live investigation and a capability building exercise for Buyer A and thus involved a study of modern slavery detection in 'real time' at a high-risk factory (Supplier). It should be noted that Supplier had experience in implementing social responsibility requirements and had previously passed their last standard social audit. However, it was identified as high risk due to it employing a large proportion of international migrant workers. It was not however a staged audit as the supplier did not know what was being investigated or how this would be done. This was the first time that they had been through a focussed assessment of their recruitment and on-site management of migrant workers. The audit took place following a modern slavery training programme, again delivered by NGO A, an experienced independent expert in this area, and consequently the team from Buyer A were well prepared. The audit team consisted of 14 people: seven from Buyer A (including one researcher, members from their UK and Asia based CR team, and the Head of Sourcing); six from NGO A (from their South East Asia team); and one from Buyer B. For the purpose of the trip and when participating in the audit (including the interviewing of workers), the researcher was introduced as a member of Buyer A. A summary of all participants is provided in Table 1, which also indicates the mnemonics used to refer to participants hereafter.
Data collection took place before, during and after the audit. The audit itself involved one of the researchers, in month 10 out of the 20-month period, travelling to South East Asia with the company's Corporate Responsibility (CR) team. The researcher was fully immersed in this trip, spending 5 full days as part of the audit team. In addition to the audit itself, the researcher also spent time with the team during the evenings allowing for more informal discussions. Informal discussions and internal meetings involving Buyer A and NGO A were audio recorded. Due to confidentiality reasons, any meetings that took place at Supplier or with members of their team were not audio recorded and instead written notes were taken. The same applies to the audit itself. Additionally, the researcher interviewed two of the supplier's migrant workers and, due to confidentiality and the sensitive nature of the discussion, these were only documented in written form rather than being audio recorded. All interviewees were selected by the audit team from a master list of employees provided by the supplier.
The final audit process given in Table 2 below summarises the data collection opportunities during the 5 day period. However, it is important to note that this audit process is also a finding of the project given that it was co-created by the audit participants and that this schedule evolved as the week progressed. Thus the researcher was very much involved in adapting it during that week, leading to the final audit protocol given in Appendix and referred to in the findings. This audit process involved the investigation of the end-to-end recruitment process using a parallel structure to provide triangulation. Thus, the audit team members were grouped into sub-teams, which were spread across three parallel sessions on Day 2 and two parallel sessions on Day 3. As shown in Table 2, the audit began on Day 1, when the management from Buyer A, NGO A and Supplier met to outline the scope and provide an overview of the schedule, ensuring that all parties understood the planned process. The main audit assessment took place over days 2 and 3 followed by a closing meeting on Day 4. Finally, the Buyer A team met on Day 5 for final reflections.
Following the audit, the researcher was involved in the remediation process for the remaining 10 months. The remediation process is ongoing and further evidence of remediation has been collected from the company since the end of the action research project.
Overall, the key data collection methods include: audit participation, observations, interviews, documentation and using diaries to record key aspects of the process for operationalising due diligence. A summary of the key data obtained is provided in Table 3. Over the 20-month period, additional dialogue took place relating to the audit during wider discussions relating to the company's response to modern slavery legislation. These discussions are all recorded in the diary entries.
Rigour has been addressed by engaging in multiple cycles of action using the framework outlined by Coughlan and Coghlan (2016). Each cycle contains a pre-step that involves understanding the rationale for action and four main steps involving constructing, planning action, taking action, and evaluating action. Monitoring and reflection has taken place throughout given that a key characteristic of action research is questioning all aspects of the research through the process of evaluation or reflection (N€ aslund, Kale, and Paulraj 2010).
The action research project can be considered one major cycle (the response to modern slavery legislation) with two minor cycles: detection and remediation (i.e. representing the main unit of analysis) taking place within the project (Coughlan and Coghlan 2016). The audit itself was a cyclical process. It was a reflective process throughout with a feedback loop built into the audit that involved a series of de-briefs between Buyer A and NGO A. This helped cross-reference findings and identify any changes to the audit process needed to follow up on key issues raised, which then fed into the next cycle. For example, the number of worker interviews was increased during the audit week after reflecting on how much information had been gained from the initial interviews. Further, this reflective practice helped to evaluate the process and develop the audit protocol presented in Appendix. Within the detection cycle there has therefore been a spiral of action research cycles (Coghlan and Brannick 2014). Findings from the detection cycle then fed into the remediation cycle. The remediation process has evolved from the initial plan and is ongoing. For example, the co-created modern slavery toolkit mentioned above emerged after reflecting on remediation-related trainingit was first a pocket guide for staff then became a smartphone application. There is therefore an ongoing spiral of action taking place and changes being made as action is evaluated.

Data analysis
The data has been coded and organised into two main themes detection and remediation. This is further divided into subthemes, which either emerged inductively from the data or deductively from the literature. Each code was discussed, and a final categorisation agreed amongst the researchers. Tabulation using these subthemes aided the analysis (Coughlan and Coghlan 2016), including the process of developing a series of vignettes, which are the primary vehicle used to describe the detection related findings in Section 4.1 below. Vignettes of this kind have been successfully employed by other authors, such as McCarthy (2017). In her study of a women's empowerment programme in Ghana, the author used vignettes to present the analysis of her findings, blending accounts from her fieldwork with reflexivity and conceptual analysis. Vignettes are described as 'stories about individuals and situations which make reference to important points in the study of perceptions, beliefs, and attitudes' (Hughes 1998: 381). They build a narrative through reflexivity for interpreting the social world (Chaudhry, 2009) and are argued to be an appropriate means of presenting the findings of this project given the importance of reflexivity when taking an action research approach (Cassell and Johnson 2006). They draw on a range of sources while researchers advocate the importance of field notes to develop personal accounts that can 'bring life to research [and] bring research to life' (Ellis 1997, 4). Thus, the vignettes presented below are written in the first person to bring the data to life and demonstrate how the audit process evolved, drawing extensively on diary notes. As encouraged by Chaudhry (2009)   As discussed above, the process of data analysis that led to the vignettes was aided by data tabulation. Initial coding therefore involved tabulation of the empirical evidence for each subtheme relating to modern slavery detection. A further three themes were later identified through studying these extensive data tables leading to a second stage of coding used to categorise the findings into the three key themes for the vignettes: (i) identifying indicators of modern slavery through hearing the voice of the worker; (ii) supplier development focus rather than monitoring; and (iii) collaboration with the NGO. Similarly, initial coding of the findings on remediation involved tabulating the empirical evidence for each subtheme relating to modern slavery remediation. A second coding step was then used to categorise the findings into the three key themes under which they are presented in Section 4.2: 'remediation actions', 'ongoing detection and remediation: engagement with local NGO' and 'ongoing detection and remediation: future audit plans'.

Findings
The findings from the detection phase are outlined in Section 4.1 using vignettes before concluding with a summary of the indicators of modern slavery in Section 4.1.1. Then, in Section 4.2, the findings related to the remediation phase are presented.

Vignette 1: Finding Clues in Worker Voices
It is morning on Day 2 of the audit and I am excited for the day ahead as I am interviewing two workers accompanied by Buyer A's UK CR Manager (BA3) and a translator from NGO A (NGOA3) who can speak the migrant workers' language. Following a debrief yesterday, we discovered that the factory is retaining the worker's bank books, so we are aware that we also need to investigate this issue. Management claimed that this was for 'safekeeping' and to facilitate the process of closing bank accounts when workers leave, but we need to further understand the issues this causes.
Although we have an idea of the key information we need to cover, each worker provides a different story, so rather than asking a set list of questions (as is common in a standard audit) we act upon leads like a detective. Later in the week I discuss this interview technique with the team. The CR Officer Asia (BA5) advises 'We learnt how to interview workers, not directly, we ask around that point e.g. [for the worker's] passport, we ask them have you seen the officer, have you taken pictures, to see if they really booked in to get a passport or someone helped them to get it' (Day 5, reflection meeting).
The first worker nervously enters the room with stiff arms down by his sides. He seems anxious and sits down, now crossing his arms. I want to put him at easehe looks cold, so I check if he is warm enough and offer him a drink. I am concerned the interview process could appear intimidating, but I am impressed with NGOA3's manner and her ability to connect with the worker. It helps that she is the same nationality as him. NGO A's Capability Building Manager (NGOA1) later explains: '[ … ] we have tried to use native speakers, if you only have 20 minutes, the barrier that you have to break if you are not even the same nationality, there will be more walls to break. That is why we always like to use native [speakers] and that is a constraint for internal auditors having that kind of resource' (NGOA1, Day 3. Post Audit Assessment Meeting).
The worker claims he is 20 years old and has worked at the factory for three years. His passport, which is in his possession, however indicates he is 22. This raises suspicion and concern about the procedures in place to screen out child labour risks. I feel a sense of sadness as he looks too young to be away from his family. I am comforted to later learn that his auntie works in the factory and his mother, sister and brother work nearby. I make a note that NGOA3 is able to provide country and regional specific data such as laws, maps, and geographic information, all necessary when dealing with migration. I later discuss with BA3 about the detailed regional specific information needed when compared to a standard audit. In particular, this was useful when the interviewees discussed their travel movements and from which location they had obtained documentation.
We move on from discussing the worker's age. He appears calmer, begins to smile and openly answers all of our questions. He enjoys working here, he loves to sew (his mother taught him when he was young) and we discussed his aspirations to become a menswear fashion designer. He explains that his parents arranged for him to migrate for work and they managed the whole recruitment process meaning he has little understanding of the details. He thinks he has a contract with a broker but does not have a copy. He does however remember his parents paying fees but he does not recall signing any documentation. The workers should not be paying fees and this is a key indicator and risk of bonded labour.
We need to establish if the workers are being exploited. Are their wages being withheld or deducted? Do they have access to disposable income? We therefore discuss how he spends his money, which is another example of a discussion that would not take place in a standard audit. He is not sure where his bank book is, but he confirms he has access to money, of which he sends a considerable sum back to his home country to help his family. It is sent via an agent; his father pays a fee at the other end. We later compare his experiences with those of other workers. Some other workers reported that without their bank books, they are unable to make bank transfers, which are more efficient and safer than using agents. I recall how earlier, at breakfast, BA3 had said how grateful she was that we would be conducting interviews together. Now I understood whyit is emotionally draining. We proceed into the second interview, but I feel tired from having to think on my feet and to quickly process the previous interviewee's responses. The second worker is older, slightly more guarded but still fairly open. I think about how nervous I would be in his position, being asked so many questions. He is married and his wife also works at the factory. He has recently renewed his contract and, although the process was quite unclear, we established that fees were deducted from his wage. His wage slip is not in his language though, so he has to get a colleague to translate it. He advises that the office holds his bank book, but he does not mind and can request it when needed. I recall our team debrief at the hotel last night when NGOA3 shared her concern that the workers do not seem to understand the risks of not being in possession of their own bank books. After the interviews, the three of us have a brief discussion. We agree that we learnt a lot from the worker interviews and discuss the importance of asking the right questions to gather the required information.
Vignette 2: Getting the Supplier On Board using a Developmental Approach It is Day 1 of the audit and prior to starting the factory investigation, we invite the Supplier management team to our hotel to meet with NGO A and plan the week ahead. We present the audit as a learning experience for both parties (Supplier and Buyer A) and it is clear that this process will be a partnership. The close relationship between the two parties is evident and the meeting is relaxed and open. The management are friendly, willing and keep smiling. Buyer A's Head of Ethical Trade (BA1) and Head of Sourcing (BA2) both stress that there is no commercial risk with regards to the audit findings and any information that is shared will be in order to enable learning, and Supplier will remain anonymous. We also discuss how the audit will inform Buyer A's policy, improve standards and enhance understanding of risk. BA1 also advises that the pilot will help build policy from the supplier perspective: 'we don't want policy that suppliers find hard to alignwe need your help, we are UK based and not manufacturing, so it is hard for us to make policy'.
The audit is presented as a general legal requirement to be transparent. Supplier appears to understand this and does not seem concerned. BA2 started the discussion regarding the audit several months ago. I later discuss this with NGO A and she explains that 'The next thing [in terms of why the audit was regarded a success by NGOA] is how Buyer A prepared Supplier [ … ] setting the tone that this is collaborative, this is working together, it is really very important but not only communicating that but taking the time to prepare the supplier for this process -that really contributed to the success' (NGOA1, Day 4, informal conversation after closing meeting).
BA1 and BA2 outline why we are doing the audit, continually stressing the strategic learning partnership. NGOA are briefly introduced as experts and BA1 explains that they have previously delivered training for Buyer A. Supplier is also advised that they are not being accused of exploiting workers but that there could be issues relating to their journey to the factory. BA1 advises that they need Supplier to be completely transparent and further stresses that there will be no negative consequences or commercial risk: 'whatever we find we will work through with you'. We move on to discuss the involvement of Buyer B who does not want to contribute financially to the audit. It has been agreed that they can observe the process on the condition that they 'engage [financially] in remediation'. I had learnt earlier in the day that they have been invited to add leverage as BA1 argued that 'There are not many factories in which we could influence policy without the support of other customers … The more leverage we have the more likely the supplier is to engage' (BA1, Day 1, informal conversation with Buyer A's audit team).
It is the following day and I am in 'Team 1' conducting interviews with Supplier's management team. Their Chief Sales Officer (S2) acts as a translator. At first Supplier's management team appear knowledgeable and do not seem to be discussing their responses with one another. But they start to make contradictory statements, particularly relating to recruitment fees. This sets alarm bells ringing and it becomes clear that the workers are being charged. Supplier's management start to become more guarded and discuss with one another before answering the questions. It begins to get more frustrating with BA2 exclaiming: 'I think they are trying to confuse us'.
Buyer B helpfully begins to translate and we start to make progress. However, BB1 becomes more confident and starts to challenge Supplier's management team. Although he asks some good questions, this raises concern as Buyer A does not want Supplier's team to stop talking. BA1 speaks to BB1 discretely ensuring that he does not become too confrontational. We do however later acknowledge that S2 would not have translated everything and we subsequently gained more information through the translation services of Buyer B.
The audit is complete and we meet as a team (Buyer A and NGOA) in the hotel to share our findings and plan our approach for the closing meeting. To avoid unnecessary resistance, we decide to focus on outlining the issues rather than framing them as human trafficking. Instead, this level of detail will be provided as the remediation plan is developed in later meetings.
It is the following day; we are sitting in Supplier's boardroom for the closing meeting to present our findings to their management team. BA2 starts by outlining areas of good practice thereby putting the supplier at ease. He follows this by outlining the issues. Supplier's CEO (S1) becomes quite defensive as the issues are explained in more detail. BA1 and BA2 adopt 'good cop, bad cop' roles and BA1 tries to relax the supplier after the issues are raised. This approach works and S1 begins to understand the issues and acknowledges that he may be unaware of what is actually happening. BA1 adds: 'If you [Supplier Management] did talk to workers at length you would find this [indicators of modern slavery] out'. He also reassures them adding 'We are not accusing [the supplier], we are learning and making you aware of the risk for you and us. You are our business partner, it is about working together, it is both our responsibility to do the right thing'.
BA1 stresses that Supplier's system is not robust enough to identify risks and that there are communication issues between workers and management. I note in my diary that the meeting is progressing well due to the approach between BA1 and BA2. BA2 takes the lead outlining the issues and BA1 tries to be positive by proposing how issues could be resolved. The rest of the meeting focuses on the next steps and outlines a remediation plan involving all parties.
The overall aim is to move away from a standard corrective action plan that leaves the supplier with a list of issues to correct. Instead, we form a collaborative working group consisting of members from Buyer A, Buyer B, and Supplier. The aim of the group is to collectively agree the supplier policy and process for future migrant recruitment. Buyer A's Head of Ethical Trade explained to Supplier: 'We think you will get more out of this from working with us, you will get to input rather than us telling youit's your business'. He also acknowledged the role of the buyer, claiming: 'We will be looking at purchasing practices to improve how the way we work with you creates problems -last minute problems, changes and no forecasts. We understand that we are involved in this process'. I make a note that this statement further emphasises the partnership approach that is being taken to make improvements.

Vignette 3: NGO A: The Backbone that Held Us Together
It is the first day at the factory and although we spent yesterday at our hotel planning with NGO A, this is our first experience of formally working together. I am in 'Team 1 0 conducting interviews with Supplier's management team and the value of having NGO A present immediately strikes me. We ask the management team some questions and request supporting documentation. I am given a number of documents to review and NGO A provide us with a helpful guide to decipher the jargon. I am sat next to NGOA2 and I ask her questions as I spot anomalies.
NGO A act as facilitator and guide us throughout the audit process, allowing us to take the lead whilst providing support and expertise. This is particularly helpful when we sense any tensions from the Supplier's management team. I later discuss this with BA3 and she claims 'I think NGO A was really good, they've got loads of expertise and they were able to facilitate it [the audit] really well and give direction. I liked that they were hands off, actually I think it was good for us, our development and use of initiative as well, not just being told what to do mindlessly' (BA3, Day 5, Reflections Meeting).
It is the following day and I am interviewing the migrant workers. Before beginning, NGOA3 starts by giving some pointers of certain issues we should cover, such as the worker's recruitment process and their understanding of their wages. Having NGOA3 present is such a great support. She knows when to push for more details and when to move on. At one point we try to establish a worker's age and she warns 'he is getting nervous'. I also make a note that NGOA3 is able to provide country and regional specific data such as laws, maps, and geographic information, all necessary when dealing with migration.
We hit a stumbling block as we try to quickly piece together the information we are being told. NGO A tells us not to worry and reassuringly advises that we can verify this in the documents later. As we uncover the anomalies, NGOA3 reassures us that 'it is very common that you find issues on the first day which you go to follow up the next day and then face new ones' She tries to put our minds at rest and explains 'Lots of these things are due to no policy and procedure so can be managed'.
We head back to the hotel and have a meeting with NGO A to help prepare for tomorrow's closing meeting. We spend time reflecting on our findings, each discussing our concerns in order to triangulate the information. At this stage we are trying to establish whether we have enough evidence to call the issue a genuine non-compliance or a risk that cannot be validated.
It is the following day and I am in the closing meeting. I accompany BA1 and BA2 and this is the only time during the audit when my presence is only to observe. We decided as a team yesterday that it was important not to overwhelm the supplier and as a result the rest of Buyer A's team remain in the hotel. NGO A provide support throughout the sometimes tense meeting, As the issues are revealed, the supplier becomes defensive at which point NGOA1 whispers to BA2 and prompts him to move to the next issue. At times BA1 asks NGO A to provide more clarification. I note that the presence of NGO A, an impartial actor, helps to provide additional leverage and credibility, particularly when the supplier acts defensively. We later discuss this closing meeting: 'The other good thing about it, what we did today, is we could say -this isn't just us, this is NGO A and us agreeing, and they are not us, they are independent experts' (BA1, Day 3, informal conversation with BA3 and BA7 after post-audit assessment meeting).
I later discuss the involvement of NGO A with Buyer A's team. We agree they were particularly helpful as we reflected throughout the process. We, for example, had debriefs throughout where we would each discuss our findings and work out a plan for the next day. BA1 argues: 'I think we shouldn't underestimate how hard we would have found it [conducting the audit] on our own'. BA3 adds: 'The findings would have been 'hard to swallow' if they [NGO A] hadn't been there' (BA3, Day 3, informal conversation with BA1 and BA7 after post-audit assessment meeting).

Audit findingsa summary
Through the triangulation of information, the audit identified 15 issues, 7 of which were categorised as 'inadequate procedures to detect or avoid modern slavery', and 8 of which were concluded to be 'indicators of modern slavery'. These issues are summarised in Table 4, although the full detailed list of indicators is not provided for confidentiality reasons.
Thus, for example, Table 4 indicates evidence was found that Supplier's current procedures for ensuring there are no document discrepancies were inadequate. BA1 confirmed that: 'Basic policies [are] in place but procedures are inadequate and not operationalised'. Further, there were issues concerning the overall transparency of the end-to-end recruitment process, particularly relating to the use of agents for recruiting migrant workers. Gaps in policy, documented processes, written procedures and their implementation were therefore deemed to result in a risk of forced labour and trafficking. Although there was no evidence of under-age workers on site, there was also concern that inadequate procedures were in place to screen out risks relating to child labour such as ensuring that all supplied documentation was authentic. Audit findings also revealed that workers paid recruitment fees in excess of legal limits and procedures to verify the fees paid by workers (via recruitment agents) were inadequate.

Remediation
In light of the above, the following subsections outline the remediation that has subsequently taken place. The actions are first summarised in Section 4.2.1 before sections 4.2.2 and 4.2.3 describe key ongoing actions relating to NGO collaboration and future audits in more detail.

Remediation actions
Buyer A took the initiative to work with Supplier, but unfortunately Buyer B did not engage as had initially been expected in the closing meeting. It became evident that Buyer B lags behind in its approach to sustainability as confirmed by BA1: '[Buyer B] is quite immature in terms of ethical trade'. As a result, the working group (as outlined in Vignette 2 above) did not continue. Instead, in response to the audit findings, Buyer A took the following action: Introduced new policies and provided guidance, to provide suppliers (including Supplier) with greater clarity on Buyer A's expectations for managing vulnerable and migrant workers; Further developed their internal training by introducing a Modern Slavery Toolkit to develop the skills of employees, particularly those that are supplier-facing, relating to spotting the signs of modern slavery; Partnered with a local South East Asia based NGO (NGO B) who has helped to embed best practices in regional suppliers; Ran a seminar introducing the new policies and NGO B, which was attended by all of Buyer A's South East Asian suppliers; Improved the worker voice through NGO B educating workers on their rights and enabling grievance reporting via the introduction of a smartphone (whistleblowing) application.
As a result, Supplier worked closely with Buyer A and NGO B to establish a plan and made significant improvements, such as: Compensating workers that had paid recruitment fees; Ceasing to charge workers recruitment fees; Updating policies and internal practices relating to employment; Improving transparency with third-party recruitment agencies; and, Listening to reports from workers via the smartphone application and responding accordingly in collaboration with NGO B.

Ongoing detection and remediation: Engagement with a local NGO
Although Buyer A initially partnered with a large multinational NGO for the targeted audit (NGO A), a local actor for the remediation stage encouraged a 'bottom up' approach (NGO B). BA1 argued that working with NGO B is helpful as: 'We don't have our own people on the ground'. For an annual fee, the collaboration with NGO B aims to improve practices by helping with policy development and implementation in their South East Asia supply chain. To achieve this, they work closely with their suppliers, workers, and thirdparty recruitment agencies on a daily basis without the involvement of Buyer A. Instead, Buyer A is sent a bi-annual report and alerted if a supplier does not engage with the NGO. However, Buyer A is not informed if there is a noncompliance, instead NGO B works collaboratively with the supplier to resolve the matter. Time is being spent educating workers so they understand their rights and are therefore able to recognise any issues themselves. Progress is monitored by giving workers a voice at all of Buyer A's suppliers in South East Asia using the aforementioned smartphone app. The app can also be used by prospective employees to learn about suppliers and recruitment agencies. NGO B combines this with conducting supplier risk assessments and interviews workers to build a picture of what is happening on a day-to-day basis. The information is then shared with the supplier so that they can work together to resolve any problems.

Ongoing detection and remediation: future audit plans
The key findings from the targeted audit are helping to shape future audits and establish an audit protocol (see Appendix). The protocol provided in Appendix was used as a starting point by Buyer A's Asia based CR team to further develop it for their own adoption in future audits. The Asia based CR team work closely with all suppliers in the region and take ownership of supplier development. The protocol is a record of what took place during the audit and can be used to replicate (and adapt) the audit to conduct targeted investigations in other suppliers and countries as required. Despite this, there is not an immediate plan to roll out the targeted audit across the supply chain. One of the main He also argued that this has not only helped focus the business's efforts but has also led to more funds being made available as the board are now confident that once issues have been found they can be remediated. The co-created modern slavery toolkit has involved using the audit findings to develop an internal training programme for all levels of employees in all departments. Supplier-facing employees are also provided with a printed pocket guide (later developed into an app) to help them spot the signs of modern slavery during supplier visits, and this provides information on worker appearance, behaviour and supplier working conditions. It is anticipated that this will be particularly beneficial to designers and buyers who regularly visit factories but whose role may not normally lead them to engage directly with social compliance.

Discussion
This study advances our knowledge on both modern slavery detection and remediation. Firstly, in terms of detection, the findings suggest a targeted audit can identify key modern slavery indicators. It is however important to acknowledge that the process may still have flaws and, in light of the current literature (Egels-Zand en 2007; Plambeck and Taylor 2016;LeBaron, Lister, and Dauvergne 2017), there is for example no guarantee that workers were not coached to reveal only the indicators management were comfortable with (and not worse atrocities). Nonetheless, the findings suggest this approach can overcome at least some of the limitations of a standard audit. It focuses on the end-to-end recruitment process, including reviewing third-party recruitment agency practices, migrant working conditions, and end of contract repatriation, addressing some of the drawbacks of the 'one-size-fits-all' generic audit referred to by New (2015). The findings provide empirical evidence of the procedure followed, addressing the call to further understand the process for a targeted audit (Gold, Trautrims, and Trodd 2015). Supporting Barrientos (2008), the imbalance of power amongst the parties involved has also been considered when developing and analysing the practices implemented.
The results have also highlighted the importance of collaboration with external parties (Buyer B, NGO A) during the detection process (Plambeck and Taylor 2016;Gold, Trautrims, and Trodd 2015), which has facilitated resource sharing, increased leverage, and influenced the supplier's level of transparency and cooperation. The roles of power and control were however evident amongst the various stakeholders. Collaborating with another buyer can provide leverage but there can be power imbalances that potentially impact the openness of the supplier. However, an NGO can act as mediator, as a support mechanism, and can also provide further leverage (Barrientos and Smith 2007). Their involvement proved particularly useful for investigating a new area such as modern slavery. In particular, their use of regional managers and local translators helped to provide a local view and awareness of the issues. This also helped to improve the diversity of the audit team.
Although the targeted audit identified modern slavery indicators, the findings also reveal the resource constraints impacting its scalability, which has led to Buyer A exploring other effective ways to detect and remediate modern slavery in the long-term. Our study provides empirical evidence of modern slavery remediation to support and extend the findings from secondary data analysis presented by Stevenson and Cole (2018). Firstly, it is important to note that following the identification of key indicators of modern slavery, Buyer A did not terminate business with the supplier, which has been identified as a response to issues in the SSCM literature (Jiang 2009;New 2015). Supporting Gold, Trautrims, and Trodd (2015), this was presumed to have a detrimental impact on the supplier, workers, and wider community. In fact, prior to the audit, Buyer A stressed to the supplier that non-compliance would not result in the withdrawal of business. Further, Buyer A's dependence on their supplier may arguably have impacted this approach (Awaysheh and Klassen 2010), and removing this risk encouraged the supplier's commitment and transparency during the audit.
The findings demonstrate the monitoring versus development approach and suggest that a more collaborative partnership can be used to detect and remediate modern slavery issues (Jiang 2009;Gold, Trautrims, and Trodd 2015). In fact, the supplier played a more active role than has previously been identified in the literature (Gould (2005;LeBaron, Lister, and Dauvergne 2017). This was evident from the early stages, including informing the supplier of audit plans several months in advance and taking the time to have a pre-assessment meeting for the management teams. This meeting took place at Buyer A's hotel which provided a more relaxed, informal environment to put the supplier at ease. From the beginning, the power dynamics at play were softened to encourage working together, ensuring policy was collectively agreed rather than the buyer telling the supplier what to do (Lund-Thomsen and Lindgreen 2014). There was also acknowledgment from Buyer A that they were not a manufacturer, making it difficult for them to determine policy. Input was therefore needed from Supplier to ensure that it was achievable. The involvement of Buyer A's Asia based CR team was also important as including the regional team improved diversity, ensuring that Buyer A's audit team was not comprised only of Westerners. Their local experience was invaluable and they were empowered to develop the future audit protocol and solutions that would be applied in the local context. Buyer A's remediation plan initially involved a collaborative working group consisting of members of Supplier, Buyer A, and Buyer B, to operationalise a corrective action plan. However, a longer-term plan has been put in place involving a local NGO (NGO B) taking the lead to manage a 'bottom-up' day-to-day remediation process helping to implement policy through supplier development. This is an alternative solution to the standard 'improvement notice', which has been questioned for its suitability during modern slavery remediation by New (2015). This approach is a move away from the compliance model focussed on monitoring that has been criticised in the literature for encouraging mock compliance (Jiang 2009;Plambeck and Taylor 2016;Huq and Stevenson 2020). Further, this replaces the 'follow-up audit' that has been identified in the prior literature as standard practice after detecting non-compliance (Jiang 2009;Stevenson and Cole 2018). Power has therefore been transferred to NGO B and Supplier to collectively resolve any issues with minimal involvement from Buyer A. This is therefore a move away from the paternalistic behaviour often demonstrated by powerful buyers (Barrientos and Smith 2007;Lund-Thomsen and Lindgreen 2014;Ozkazanc-Pan 2019).
Involving a local NGO in supplier development can therefore be seen as an alternative measure to eliminate the need to re-audit and reduce audit fatigue (Awaysheh and Klassen 2010;Marshall et al. 2016). The findings therefore also add to the limited research on collaboration with NGOs from a social sustainability perspective (Zorzini et al. 2015) and provide further examples of their involvement in both worker and supplier development, thereby supporting Huq, Chowdhury, and Klassen (2016). This has involved working with suppliers to develop documentation systems, policies, and procedures. In particular, the findings reveal the importance of educating workers on their rights and whistleblowing achieved through the use of a smartphone application. This helps to reduce the risk of deception that can take place during audits if workers are coached or feel under threat from their employers if they raise concerns (Egels-Zand en 2007; Plambeck and Taylor 2016). NGO B is also working closely with recruitment agencies to develop an ethical recruitment process, which provides an example of how a company can tackle exploitative third-party recruitment agencies (Barrientos, Kothari, and Phillips 2013). As a result, NGO B is able to detect, mitigate, and remediate indicators of modern slavery on a daily basis through building close relationships with workers, suppliers, and recruitment agencies. Thus the findings provide evidence of how 'workerdriven monitoring' can be achieved in practice, as advocated by authors such as Prieto-Carr on et al. (2006) and Outhwaite and Martin-Ortega (2019). Empowering workers in this manner has not only involved them being trained to further understand their rights but ensures that they are actively involved in the detection and remediation of modern slavery. This helps to ensure that decisions are not solely made by a more powerful party (e.g. a buyer or supplier), and their view of what is in the workers' best interest (Barrientos and Smith 2007).
It should also be noted that investigating modern slavery is 'new territory' for both brands and suppliers. It is therefore anticipated that, in time, suppliers will understand which processes will be investigated, which could arguably lead to the falsification of documentation, and so onleading to mock compliance and hiding violations, as evidenced in the wider SSCM literature (e.g. Huq, Stevenson, and Zorzini 2014;Huq and Stevenson 2020). Having a local NGO 'on the ground' could however help to minimise this through improving transparency, training workers, and developing suppliers/recruitment agencies in sustainable practices.
Evidence also reveals action that has taken place to remediate the specific issues found during the audit relating to fees, such as paying compensation to workers. Further, the audit has led to Buyer A seriously considering the impact of their commercial power (Gereffi 1999;Anner 2019) and consequently beginning the process of making internal changes to their purchasing practices (Barrientos 2008;New 2015) as well as improving company-wide modern slavery awareness through training and the development of a modern slavery toolkit. Modern slavery awareness is therefore being embedded into the everyday practices (Pagell and Shevchenko 2014). Further, the findings reveal that when the detection and remediation process is perceived to be effective, this can strengthen top management support for integrating modern slavery awareness within the business (Pagell and Wu 2009). In this research, the CR team of Buyer A felt they had convinced their board of directors that, by conducting a targeted audit, they understand the risks within their supply chain and can remediate effectively.
Modern slavery is by no means simple to investigate given its criminal nature. For Buyer A, this has been a learning experience. Conducting a targeted audit at a high risk factory was a logical starting point which has subsequently led to the development of their 'bottom up' worker-driven approach. The audit has also helped to inform training and the modern slavery toolkit as well as understanding what steps needed to be taken for the whole supply chain. The targeted audit has therefore played an important role in both the detection and remediation process. Yet, the effectiveness of it has been underpinned by many factors such as being able to hear the voice of the workers, the involvement of multiple stakeholders and the subsequent power, control and trust at play. The initial audit suggested that the workers were too trusting of their employer, given that they were not uneasy about their bank books being held by the office. However, subsequent training of the workers through the involvement of the local NGO can better aid workers in understanding their rights thereby making them less vulnerable to exploitation and help them to be involved in the detection and remediation process.

Conclusions
This paper started by asking 'How can audits be improved to better detect modern slavery in the supply chain, and how can an appropriate remediation plan be established when modern slavery is discovered?' Action research has been used within the textiles and fashion industry to understand an attempt at modern slavery detection and the development of a remediation plan. This method has enabled the gathering of rich qualitative data concerning a sensitive high-profile issue, further advocating the use of innovative research methods for researching SSCM (Touboulic and Walker 2016).
To the best of our knowledge, this is the first empirical paper to consider how modern slavery is being investigated and remediated within a supplier factory. Whilst prior literature has highlighted the shortcomings of supply chain auditing and questioned the suitability of this approach for detecting modern slavery, this paper suggests that a more targeted audit can identify key indicators of modern slavery. This type of audit includes investigating the end-to-end recruitment process by using a parallel structure of management and worker interviews and documentation review. This has also included the investigation of third-party recruitment agency practices. Further, the audit involved collaborating with another buyer and an NGO to add leverage, share resources and develop capabilities. Although the targeted audit is resource intensive and therefore its scalability can be questioned, it is argued that companies could initially engage in this level of investigation to develop their longterm plans for modern slavery detection and remediation.
The research also documents the evolving and ongoing remediation process, which has led to collaboration with a local NGO to support workers and develop suppliers. This provides an alternative approach to remediation that simultaneously provides continual detection of the indicators of modern slavery by focussing on day-to-day collaboration with suppliers and workers, which in turn reduces reliance on resource-intensive audits.
Given the issues of trust, power and control that can be at play when investigating issues of a criminal nature, such as modern slavery, this research has highlighted the need to find appropriate ways to hear the worker voice during both the initial audit and the ongoing detection and remediation. Their voice is an essential part of understanding the reality of their working environment. It is therefore important to overcome the power asymmetries between the buyer, supplier, NGO and the workers. By using a collaborative, developmental approach with the supplier, the audit and ongoing remediation has fostered trust between buyer and supplier, which in turn fosters the supplier's willingness for workers to be involved in the process of detecting and then improving standards. Whilst other studies have drawn attention to the potential power of the buyer over the NGO, this study suggests that a collaborative partnership between buyers and NGOs is also possible, and is more conducive to combatting the modern slavery threat.

Implications for practice
This paper will be of benefit to managers, particularly in the textiles and fashion industry, aiding the improvement of social sustainability in the context of recent modern slavery legislation. The findings provide managers with a protocol for designing a targeted modern slavery audit and developing a remediation plan, which is argued to be different from the practices employed for other social issues due to the hidden and criminal nature and the requirement to investigate the end-to-end recruitment process, including third-party recruitment agencies. Conducting a targeted audit in a highrisk supplier is encouraged to help the business understand risk, which will in turn help develop a process to detect and remediate modern slavery. However, it is noted that this may not be feasible on a larger scale.
A targeted audit can be facilitated by collaboration with other buyers or non-business actors such as NGOs to help provide mediation, leverage, share resources, and develop knowledge and expertise. In particular, collaboration with other buyers can help to address resource intensity and develop a cohesive approach to resolving issues such as modern slavery. It is also important for buyers to develop an equal relationship with their suppliers, ensuring that a buyer does not assume a more powerful position and exert pressure on suppliers to comply with policy that has only been developed from the buyer's perspective. Meanwhile, enlisting the help of NGOs who are experts in modern slavery investigation can help to verify risks and provide country and regional specific data such as laws and geographic information, which is key when investigating migrant labour. It may however not always be feasible to have an NGO and/or native speaker present. There may therefore be practical difficulties due to being unable to speak to workers in their own language and build a rapport.
Companies should also consider collaborating with local partners, including local NGOs, to facilitate 'bottom-up' detection and remediation by providing day-to-day 'on the ground' support through developing and empowering workers and suppliers. Whistleblowing is encouraged by working closely with workers and educating them on their rights so they are able to request advice, report issues and assume a more active role in the detection and remediation process. This can be facilitated using an NGO-operated smartphone application, providing the NGO with the means to verify issues and resolve them with the supplier without alerting the buyer.
It is also important for all factory-facing employees to understand the causes and signs of modern slavery. Buyers are therefore encouraged to focus on their internal processes and embed modern slavery awareness throughout their company by improving purchasing practices and transparency, and by providing internal training and resources to support employees at all levels of the business. Developing a collaborative approach in the detection and remediation of modern slavery is meaningless if the buyer's internal practices are not continually reviewed.

Limitations and future research
In light of the prior literature, the targeted audit process could still have flaws. Firstly, the master list of employees, from which we selected interviewees, may not have included everyone. Additionally, there is of course no guarantee that the workers were not coached on what to say. The research is also limited in terms of its supply chain scope given that it investigated modern slavery detection and remediation for Buyer A and one of their tier-one suppliers. Further investigation could be conducted of product assembly that takes place outside the factory walls and beyond the immediate first tier, such as homeworkinga context in which many vulnerable workers are employed in the textile industry in developing countries. The research could also be extended to include other lower tier materials and component suppliers. Additionally, third-party recruitment agencies could be involved in future studies. Such studies could potentially uncover new issues and challenges or require a different approach to the detection and remediation of modern slavery. This could also build on the literature that has considered how responsibility is delegated for managing sustainability amongst sub-suppliers in complex multi-tier supply chains (Wilhelm, Blome, Wieck, et al. 2016;Wilhelm, Blome, Bhakoo, et al. 2016;Grimm et al. 2016).
Additionally, research could validate and further develop the audit protocol presented. Future research could also continue to assess the remediation process by, for example, investigating the effectiveness of the smartphone application and modern slavery toolkit. For example, it will be important to further investigate how issues of trust, power and control impact their implementation. In particular, it is necessary to understand the views of more of the stakeholders involved, including the workers. Finally, as the audit and remediation processes presented here are clearly not fool-proof and the criminal perpetrators of modern slavery may become wise to these approaches, there is an ongoing need to research smarter methods of modern slavery detection and remediation.

Disclosure statement
No potential conflict of interest was reported by the author(s).

Funding
This work was supported by the Economic and Social Research Council (10.13039/501100000269ES/J500094/1).

Notes on contributors
Dr Amy V. Benstead is a Lecturer in Fashion Management in the Department of Materials, University of Manchester, UK. She holds a PhD in Management Science from Lancaster University Management School and is a member of the European Operations Management Association (EurOMA). Her research interests include socially sustainable supply chain management and global sourcing, with a particular focus on the textiles and fashion industry. She has considerable knowledge and international experience gained in the fashion industry, most recently as Senior Merchandise Manager based in Istanbul, Turkey at global sourcing company Li & Fung where her main responsibilities included managing the sourcing and production for major UK fashion retailers. Amy's work has been published in the International Journal of Operations & Production Management (IJOPM). Linda C. Hendry is a Professor of Operations Management at Lancaster University Management School, UK. Her research interests include: global supply chain management, including sustainable sourcing and supply chain responses to modern slavery legislation; manufacturing strategy, planning & control for product customisation contexts; and process improvement approaches, such as Six Sigma. Linda is Member of the European Operations Management Association (and was formerly on the Board as a member of the Finance Team). She has published extensively in a wide variety of journals, including those that focus on Operations Management, Production and Operational Research.