The Accord on Fire and Building Safety in Bangladesh: A New Paradigm for Limiting Buyers’ Liability in Global Supply Chains?

The Accord on Fire and Building Safety in Bangladesh (the Accord) is generally seen as a positive development in ensuring that Bangladeshi garment industry workers have access to safe working conditions. A central structural difference between the Accord and earlier corporate social responsibility (CSR) initiatives is that the Accord takes the form of an enforceable contract that directly connects first-world buyers with representatives of the third-world laborers of their supply chains. Traditionally, CSR mechanisms tread a fine line between a promise of decent labor conditions, often targeted at first-world consumers, and the nonbinding nature of such mechanisms, at least from the perspective of third-world laborers. The chief competitor of the Accord, the Alliance for Bangladesh Worker Safety (the Alliance), follows the traditional model. Thus the Accord represents a break from earlier nonbinding and worker-exclusive CSR by providing a new paradigm stressing enforceability and inclusivity. The novel structural aspects of the Accord are viewed positively by scholarship, interest groups, and general reporting. My starting point is this distinction between the positive, empowering image attributed to the enforceable agreement in the case of the Accord and the negative, hollow-words image of compliance mechanisms that do not take the form of an enforceable agreement, such as the Alliance. I argue that the possibilities for controlling liability allowed by an enforceable governance agreement can outweigh the possibilities for controlling liability allowed by reliance on strict conceptions of privity. From this perspective, the Accord can be critiqued as the herald of a new CSR paradigm that allows buyers new methods for controlling liability over their global supply chains. Additionally, the new paradigm comes with a whitewashing effect towards consumers and regulators. I argue that even more pronounced, however, can be its whitewashing effect towards adjudicators. Courts and arbitral tribunals may be prone to value the sanctity of the four-corners private ordering of transnational contracts, such as the Accord, over locally embedded legal safeguards.

The Accord on Fire and Building Safety in Bangladesh (the Accord) is generally seen as a positive development in ensuring that Bangladeshi garment industry workers have access to safe working conditions. A central structural difference between the Accord and earlier corporate social responsibility (CSR) initiatives is that the Accord takes the form of an enforceable contract that directly connects first-world buyers with representatives of the third-world laborers of their supply chains. Traditionally, CSR mechanisms tread a fine line between a promise of decent labor conditions, often targeted at first-world consumers, and the nonbinding nature of such mechanisms, at least from the perspective of third-world laborers. The chief competitor of the Accord, the Alliance for Bangladesh Worker Safety (the Alliance), follows the traditional model. Thus the Accord represents a break from earlier nonbinding and worker-exclusive CSR by providing a new paradigm stressing enforceability and inclusivity.
The novel structural aspects of the Accord are viewed positively by scholarship, interest groups, and general reporting. My starting point is this distinction between the positive, empowering image attributed to the enforceable agreement in the case of the Accord and the negative, hollow-words image of compliance mechanisms that do not take the form of an enforceable agreement, such as the Alliance. I argue that the possibilities for controlling liability allowed by an enforceable governance agreement can outweigh the possibilities for controlling liability allowed by reliance on strict conceptions of privity. From this 412 THE AMERICAN JOURNAL OF COMPARATIVE LAW [Vol. 66 perspective, the Accord can be critiqued as the herald of a new CSR paradigm that allows buyers new methods for controlling liability over their global supply chains. Additionally, the new paradigm comes with a whitewashing effect towards consumers and regulators. I argue that even more pronounced, however, can be its whitewashing effect towards adjudicators. Courts and arbitral tribunals may be prone to value the sanctity of the four-corners private ordering of transnational contracts, such as the Accord, over locally embedded legal safeguards.

IntroductIon: From nonbIndIng to EnForcEablE corporatE SocIal rESponSIbIlIty
The Accord on Fire and Building Safety in Bangladesh (the Accord) 1 is generally seen as a positive development in ensuring that third-world garment industry workers have access to safe working conditions. 2 A central structural difference between the Accord and most earlier corporate social responsibility (CSR) initiatives is that the Accord takes the form of an enforceable contract that provides direct benefits to workers in Bangladeshi factories. 3 Traditionally, CSR mechanisms tread a fine line between a promise of decent labor conditions, often targeted at first-world consumers, and the nonbinding nature of such mechanisms, at least from the perspective of third-world laborers. 4 The chief competitor of the Accord, the Alliance for Bangladesh Worker Safety (the Alliance), follows the traditional model. 5 The question of how to organize the governance of a supply chain, via enforceable governance contracts, such as the Accord, or nonbinding governance mechanisms, such as the Alliance, has broad consequences. Production often takes place in global supply chains consisting [Vol. 66 of their salient features and critiques. In Part II, focus is on a general critique of governance through contract. This critique is coupled with a historical-comparative look at how law has countered the liabilitylimiting effects of various kinds of contractual structures in relation to defective products, a development arguably comparable to supplychain liability or, put another way, production liability. I then extend this critique to discuss its implications for understanding the two paradigms of controlling supply-chain liability represented by the Accord and Alliance. In Part III, I look at current discussions of extending liability in global supply chains. I argue that, from a buyer's perspective, there is a tangible risk for seeing buyers as liable for damages arising out of inadequate control of their supply chains when governance mechanisms similar to the Alliance are used. In Part IV, I argue that if mechanisms such as the Alliance no longer offer adequate certainty for controlling liability, buyers may be pushed towards using direct governance contracts, such as the Accord, to control liability. I ground this argument in a discussion of the possibilities allowed by a direct governance contract for controlling supply-chain liability using the Accord as an example. In the Conclusion, I outline the differences of enforceable and inclusive CSR mechanisms, as represented by the Accord, and the existing paradigm of nonbinding and exclusive CSR mechanisms, as represented by the Alliance, and note the general potential for a paradigm shift in CSR mechanisms. Despite their radically different legal underpinnings, these two paradigms share similar fundamental flaws in their reliance on the benevolence of global buyers. This leads me to raise the question of the adequacy of existing legal safeguards if structures akin to the Accord see increased use for controlling liability.
I. tWo contraStIng approachES to cSr: thE accord on FIrE and buIldIng SaFEty In bangladESh and thE allIancE For bangladESh WorkEr SaFEty

A. Same Old Story? A Catastrophe-Turned-Media-Uproar Gives Rise to New CSR Instruments 10
On April 24, 2013, the Rana Plaza building, housing among other things several garment factories in Savar, Bangladesh, collapsed catastrophically, resulting in the deaths of over a thousand factory workers and global media uproar. 11 In the aftermath, some companies utilizing 10. For example, Nike's early CSR initiatives in the 1990s are seen as a response to outrage following media coverage of working conditions at supplier factories. See lockE, supra note 6, at 49-50.
11. For example, the New York Times has extensively covered the disaster and the different global initiatives that have followed in its aftermath. Using the search term "Rana Plaza" on the New York Times website provides a source of general reporting on the topic. For an overview, see Editorial, One Year After Rana Plaza, n.y. tImES (Apr.

2018]
Bangladeshi garment suppliers went as far as to withdraw all production from Bangladesh due to the local government's perceived failure to enforce safety standards. 12 A number of fashion and retail companies, however, attempted a more constructive approach to improving working conditions in Bangladesh. This has resulted in two competing CSR initiatives, the Accord on Fire and Building Safety in Bangladesh and the Alliance for Bangladesh Worker Safety. 13 The success of these approaches in benefiting the workers of the Bangladesh garment industry is under debate. 14 Regardless, the two initiatives offer an interesting comparison between two structurally very different models of supply-chain governance. One, the Accord, is based on a dedicated governance contract between buyer companies and the global and local representatives of their suppliers' workers. 13. See bangladESh accord, supra note 1; allIancE For bangladESh WorkEr SaFEty, supra note 5. For scholarly discussion comparing and situating the two initiatives in a broader context, see, e.g., Anner, Bair & Blasi, supra note 2 (situating the Accord in global and historical contexts, in particular in comparison to jobber's agreements in twentieth-century United States); ter Haar & Keune, supra note 2 (comparing the Accord with the Alliance from a transnational private regulation perspective); Benjamin A. Evans, Accord on Fire and Building Safety in Bangladesh: An International Response to Bangladesh Labor Conditions, 40 n.c. J. Int'l l. & com. rEg. 597 (2014) (discussing the Accord and the Alliance in light of their background); Alexandra Rose Caleca, The Effects of Globalization on Bangladesh's Ready-Made Garment Industry: The High Cost of Cheap Clothing, 40 brook. J. Int'l l. 279 (2014) (arguing for a regulatory effort in the United States, in particular via certification, for improving working conditions in the ready-made-garment sector in Bangladesh). For a comparison of the two initiatives by workers' rights activists, see Comparison: The Accord on Fire and Building Safety in Bangladesh and the Gap/Walmart Scheme, clEan clothES campaIgn (July 10, 2013), http://www.cleanclothes.org/resources/ background/comparison-safety-accord-and-the-gap-walmart-scheme/view. 14. For a critical view claiming that only a minority of Bangladeshi workers are covered by the two initiatives, see Sarah laboWItz & dorothéE baumann-pauly, n.y.u. StErn cEntEr For buSInESS and human rIghtS, bEyond thE tIp oF thE IcEbErg: bangladESh'S ForgottEn apparEl WorkErS (2015). For an opposite view critiquing the methodology of the previous study and arguing that the majority of Bangladesh garment workers are in fact covered, see mark annEr & JEnnIFEr baIr, pEnn StatE cEntrE For  [Vol. 66 The other, the Alliance, is based on a group of buyer companies pooling together to coordinate gratuitous aid without an agreement enforceable by the representatives of their suppliers' workers. To facilitate discussion of these two models of supply-chain governance and the paradigm shift in governance that they may entail, I outline in the following two subsections some salient features, differences, and critiques of the Accord and the Alliance.

B. The Accord
The website of the Accord describes it as "an independent, legally binding agreement between brands and trade unions designed to work towards a safe and healthy Bangladeshi Ready-Made Garment Industry." 15 This description highlights the two novel structural characteristics of the Accord that distinguish it from other CSR initiatives. The first is the nature of the Accord as an "independent, legally binding agreement." The direct enforceability of the Accord via its arbitration clause is a major differentiation from most, if not all, other private CSR alternatives. It has also been a deal-killing issue for a number of U.S. buyers that have refused to join the Accord due to its binding nature and thus also a contributory reason for the existence of two initiatives, the Accord and the Alliance, instead of just one. 16 The second defining structural characteristic is the inclusion of not just first-world buyers but also of global and Bangladeshi trade unions as parties to the agreement. For example, even when Beryl ter Haar and Maarten Keune argue that as an example of transnational private regulation (TPR) the Accord cannot create de jure legally binding norms, the inclusion of worker representatives is an important legitimating factor for any de facto legal effects the Accord may have. 17 The inclusivity of the Accord has also historical significance. Mark Anner, Jennifer Bair, and Jeremy Blasi view the Accord in this sense as a likeness to twentieth-century jobber's agreements in the United States, through which trade unions contractually arranged what in practice was joint liability between contractors and employers towards employees in the garment industry. 18 Its nature as a binding and enforceable agreement and its inclusivity of supplier employees via their representatives are the foundational structural characteristics of the Accord, differentiating it from earlier TPR/CSR instruments. 19 In practice, these two characteristics translate to a number of more specific contractual provisions. Key

2018]
provisions that the buyer signatories commit to over the five-year span of the Accord include the following: • disclosure of Bangladeshi suppliers and submitting these suppliers to safety inspections led by recognized and independent experts; 20 • public disclosure of inspection reports; 21 • requiring suppliers to implement repairs and renovations as determined in the inspections; 22 • paying suppliers prices that make possible repairs and renovations and the safe operation of factories and maintaining supplier relationships throughout the five-year length of the program; 23 • allowing democratically elected worker representatives into supplier factories to educate workers about workplace safety and worker rights; 24 • guaranteeing certain rights to workers of factories affected by renovations or by the implementation of the Accord, such as the right to refuse unsafe work conditions and efforts towards maintaining their income during factory downtime; 25  25. See bangladESh accord, supra note 1, arts. 13-15. The more direct workerrelated obligations of the signatories include requiring suppliers to maintain workers' employment and income during any downtime for renovations for up to six months (Article 13), reasonable efforts to find safe work places for those whose employment has been terminated due to a loss of orders (Article 14); and requiring suppliers to respect workers' rights to refuse unsafe work without discrimination or loss of pay (Article 15). These are discussed in more detail in Part IV. 26 Despite its novel approach in the global TPR/CSR context, the Accord has also been criticized. A major focus of this critique is the narrow subjective scope of the Accord. For example Anner, Bair, and Blasi, who compare the Accord to earlier U.S. jobber's agreements, note that the Accord "covers only one area of labor standards-worker safety-which was not the focus of the jobber's agreements," while jobber's agreements were more broadly oriented toward joint liability of buyers and suppliers towards workers for fair labor conditions. 27 The efficiency of the Accord in providing Bangladeshi workers with safe working places is also under critical debate. 28 A critical assessment of the overall efficiency of the Accord, when compared to the Alliance, may be necessary in evaluating whether the structural advances of the Accord can in practice make a positive difference in worker safety when compared to other initiatives, but that discussion is beyond the scope of this Article.
The enforcement mechanism included in the Accord in any case seems to work. In two arbitrations raised by union parties against global fashion brands who were not seen to act in accordance with the Accord, the procedural objections made by the garment companies were overcome following which they agreed to settle. 29 Furthermore, after the upcoming expiration of the five-year term of the Accord in May 2018, a successor, dubbed the "2018 Accord on Fire and Building Safety in Bangladesh," is set to enter into force on May 31, 2018. 30 While generally similar in substance, the 2018 Accord does contain some changes, for example, in relation to dispute resolution. 31 In short, the Accord is a dedicated governance contract that connects the two ends of a global supply chain. 32 Within the limited field of fire and building safety in the Bangladeshi garment industry, the Accord complements the chains of contracts that constitute a supply chain between first-world buyers and Bangladeshi suppliers. While the Accord as a governance contract does not directly

2018]
affect the relationship between suppliers and their workers due to the suppliers not being party to the Accord, it creates a direct contractual link between buyer companies and the representatives of their suppliers' employees. This raises the question of what the buyers receive from the agreement from a legal standing point: Is the direct contractual relationship used not only to coordinate certain benefits and rights to supplier employees but also to control the liability of the buyers?

C. The Alliance
Unlike the Accord, which is an agreement between buyers and unions representing supplier employees, the Alliance for Bangladesh Worker Safety is a Delaware corporation with primarily North American buyers as its members. The Alliance uses similar language in its marketing as the Accord, for example by describing itself as: "a legally binding, five-year commitment to improve safety in Bangladeshi ready-made garment (RMG) factories." 33 So what is the difference between the legally binding commitment, as per the Alliance, and the legally binding agreement, as per the Accord? According to the Alliance FAQ, "legally binding" in the case of the Alliance is understood as follows: Membership to the Alliance is a five-year commitment-and the agreement and its terms that are legally binding on all of its Members. The Alliance Board of Directors-chaired by an Independent Director-has the authority to seek binding arbitration against any Member who does not satisfy its obligations under the agreement, and to publicly expel them for failure to abide by other commitments set forth in the Members Agreement. 34 This approach has been criticized by workers' rights groups who argue that the Alliance superficially resembles the Accord while omitting the Accord's most important substantive elements. 35  [Vol. 66 contributions from participants apart from administrative fees and a nonobligatory loan program; second, that the Alliance lacks enforcement provisions; and third, that the Alliance was developed and is governed without worker participation. According to workers' rights activists, this leads to the Alliance working to "undercut the Accord by providing a less onerous and less rigorous alternative." 37 Ter Haar and Keune offer a similar critique of the legitimacy of the Alliance due to the exclusion of worker representatives from any binding mechanism. 38 In particular, they note that the Alliance relies on traditional command-and-control mechanisms and monitoring by means of financial-style auditing, with little attention for the capabilities of the Bangladeshi factories and workers to identify and address problems, or the dangers of unreliable or false information supplied to audits. As such, it bears a strong resemblance to the early period of unilateral CSR codes that were often not effective in practice. 39 Finally, the critique aimed at the Accord's narrow scope, discussed above in relation to the Accord, is similarly applicable to the Alliance. As with the Accord, whether the Alliance is empty promises or making a change is under debate. 40 Furthermore, the Accord and the Alliance exist at least partially in mutual recognition of one another, in particular due to the fact that supplier factories may produce clothing for both Accord parties and Alliance members and therefore may fall under the scope of both initiatives. 41 Again, the relative success of either initiative is not the focus here. Instead, my focus is on the different approaches to controlling liability that the two initiatives entail. From this perspective, the differences of the two in relation to the contractual construction of supply chain governance are crucial. The Accord is an agreement between buyers and the representatives of supplier employees to coordinate factory safety, whereas the Alliance is a commitment between buyers to coordinate safety efforts aimed at supplier factories. To make the difference between the two even clearer, the Alliance includes explicit noncommittal language that is lacking in the Accord.

2018]
The sole rights to enforce any alleged breach of such commitments by a Member are through the processes described in the Bylaws. No Member has any right of action or other claim against another Member arising out of this Agreement, or such Member's participation in the Alliance, all of which are hereby waived and released. 42 From the perspective of this Article the crucial difference between the Alliance and the Accord is that the Alliance specifically tries not to establish a legally relevant connection between buyers and their suppliers' employees. Instead, Alliance members try to avoid any appearance of such a relationship and resort to the lack of privity between them and their suppliers' employees apparently as a means of controlling liability. The Accord, on the contrary, establishes privity between Accord companies and the representatives of their supplier employees via a direct contract. This difference between the Alliance and the Accord has a crucial impact on how liability may be conceptualized in the case of either governance mechanism. To understand the effects of these two radically different approaches to controlling liability between buyers and their suppliers' employees, I will next look at the relationship between contractual governance mechanisms and liability.

A. Systematizing the Contractual Governance of Supply Chains
Contract governance is a broad interdisciplinary topic focusing on the role and regulation of contracts in governance. 43 Florian Möslein and Karl Riesenhuber see it as comprising four potentially overlapping topics of research. 44 The first topic is "governance of contract law," with a focus on the institutional framework of contract law rule making. 45 The second topic is "governance of contracts," with a focus on institutions that constitute the framework of private transactions, such as pacta sunt servanda and its various exceptions. 46 The third topic is "governance by means of contract law," where the focus is on using contract law as an instrument to achieve specific regulatory 42 goals, similarly to using tax law in steering habits of consumption. 47 The fourth topic is "governance through contract," where the focus is on using contracts to create frameworks of governance. 48 Following Möslein and Riesenhuber's framework, my focus here is on the disjuncture between governance of contracts and governance through contract. Peer Zumbansen has discussed the relationship of the two under American legal theory. 49 In particular, in some cases, governance through contract can be used to configure legal relationships in a way that overrides the ability of governance of contracts to balance the interests of the different parties. 50 In relation to supply chains, this disjuncture can extend to relationships between actors that are not party to the same contract. For example, the Alliance is a governance mechanism that allows buyers control over their supply chains without a direct contractual relationship that extends to all other actors in the supply chain. Similarly, codes of conduct incorporated in the contract between a buyer and its first-tier supplier may be intended to affect the whole supply chain even when the buyer is in privity only with its first-tier supplier.
Legal scholarship has not focused on separating different types of governance through contract. 51 Other disciplines have more readily taken up the task of developing typological models of governance through contract. A number of these approaches build on Ronald Coase's classic question of why actors in some cases choose to govern their supply relationships through contracts relying on marketprice mechanisms ("markets") while in other cases they choose to vertically integrate supply relationships into hierarchically governed corporations ("hierarchies"). 52 In particular, research has focused on whether there are further modes of organization beyond markets and hierarchies. 53  51. While different contract types are clearly regulated by different sets of rules, discussion related to governance through contract seems to focus more generally on the use of contracts to remove private ordering from public regulation. See, e.g., Zumbansen, supra note 43, at 184-85; Zumbansen, supra note 49. Some exceptions may be provided by multidisciplinary work grounded in both social sciences and legal dogmatics, such as for example Gunther Teubner's notion of networked contracts, where under specific factual conditions special relationships (instead of contractual relationships) may arise between actors not connected by a formal contract but involved in the same network of contracts.

2018]
of organization based on contract include Oliver Williamson's 1979 paper Transaction-Cost Economics: The Governance of Contractual Relations 54 and Gary Gereffi, John Humphrey, and Timothy Sturgeon's 2005 paper The Governance of Global Value Chains. 55 Both Williamson and Gereffi, Humphrey, and Sturgeon argue that the special characteristics of certain supply relationships result in modes of governance that are different from both markets and hierarchies. 56 In both cases these new modes of governance are based on contract. Williamson identifies two additional types of governance through contract between markets and hierarchies, 57 while Gereffi and colleagues identify three additional types of governance through contract. 58 From a legal perspective, both models can be critiqued for their theoretical and practical vagueness. 59 Furthermore, from the perspective of supplychain governance, a key challenge is that existing literature focuses primarily on bilateral contractual relationships, i.e., a traditional notion of the parties to a contract. 60  56. For Williamson, key factors in locating these additional forms of governance are asset-specificity, recurrence, and uncertainty, while for Gereffi, Humphrey, and Sturgeon, the focus is on the complexity of transactions, the ability of actors to codify transactions, and the capabilities of the supply base. See Williamson, supra note 54, at 239; Gereffi, Humphrey & Sturgeon, supra note 55, at 85. 57. The additional forms of governance through contract identified by Williamson are "trilateral governance," focusing in particular on the possibilities afforded by thirdparty-led dispute resolution and standards, and "bilateral governance," encompassing various "relational" mechanisms for adapting contracts. See Williamson, supra note 54, at 247-53.
58. The additional forms of governance through contract identified by Gereffi, Humphrey, and Sturgeon are "modular governance," based on the use of standards, "relational governance," based on either close relationships of trust or specific contractual mechanisms, and "captive governance," which is characterized by asymmetrical power relationships resulting in contractual practices that, from a legal standing point, could perhaps be categorized as unconscionable or near-unconscionable. Gereffi, Humphrey & Sturgeon, supra note 55, at 84-88.
59. For example, Williamson, supra note 54, at 238 n.26, is right in noting the importance of generalized models of contract for theory. Nonetheless, Williamson's model becomes vague in part due to its use of Ian Macneil's theorization of relational contract, which is vague in itself. 60. For example, Möslein & Riesenhuber, supra note 43, at 284, state that under governance through contract "the range of governance is different, given that governance through contract can, due to the privity of contract, only affect the relationship of the parties as such." While they acknowledge that contracts can have effects beyond the parties through enforceable promises to third parties, they do not discuss the effects that contractual private ordering may have on other forms of action, such as founding and limiting third party claims arising in tort, which would be important for a broader systematization of third-party effects.
[Vol. 66 Humphrey, and Sturgeon, which is ostensibly focused on supply chains, primarily deals with the governance aspects of the relationship between a lead firm and its first-tier supplier. 61 In formulating a legally meaningful approach to the contractual governance of supply chains, this could be seen as a serious flaw: it is generally acknowledged that contracts can extend their reach to nonparties, for example by founding and limiting tortious causes of action from third parties. 62 Empirical studies have also described the use of contractual mechanisms to extend governance beyond privity. 63 In particular, Richard Locke's extensive work focuses on the practical effect that different governance through contract approaches implemented by buyers have on supplier employees. 64 Locke identifies three consecutive stages of development in supply-chain governance in relation to governing working conditions in supplier factories, each stage trying to overcome its predecessor's defects but simultaneously laden with its own shortcomings. 65 This leads Locke to critique the general inadequacy of current modes of governance through contract in global supply chains. On the one hand, he argues that buyers should be more deeply integrated into their supply chains, for example through partnering. 66 61. Gereffi, Humphrey & Sturgeon, supra note 55, at 98, state that the "global value chains framework focuses on the nature and content of the inter-firm linkages, and the power that regulates value chain coordination, mainly between buyers and the first few tiers of suppliers [as opposed to a value-chain-wide-governance perspective taking into account both ends of a value chain]." While this would imply a focus on at least a few tiers of suppliers, in practice the provided examples center on bilateral relationships, as does the graphical overview of different nonhierarchical governance types presented in Id. at 89 ( fig.1

2018]
On the other hand, he highlights the necessity of public regulation in complementing governance through contract. 67 Together, these two critiques could be seen to comprise a fourth stage of development in relation to supply-chain governance. Locke's work offers an interesting contrast to the models of Williamson and Gereffi and colleagues. While all three critiques focus on contracts as governance structures, Locke is the only one to conduct an extensive study focusing on actual contractual provisions and their effects. Thus, he offers a more robust starting point for differentiating between forms of governance through contract. Not only that, but he also offers an empirically backed critique of governance through contract by providing concrete evidence of its reliance on either the benevolence of buyers or the efficiency of regulators.
In line with their nonlegal approaches, none of these perspectives on supply-chain governance discusses the legal effects of different governance mechanisms. To remedy this, I have proposed earlier a more legally focused framework of "contract-boundary-spanning" governance mechanisms. 68 In the current preliminary stage of the framework, I attempt to overcome the vagueness (from a contractual perspective) of Gereffi, Humphrey, and Sturgeon's work by combining it with empirical research on contractual mechanisms, such as Locke's work, in order to provide an analytical tool that is more descriptive from the perspective of legal doctrines. In particular, the framework focuses on different contractual methods for extending control beyond privity. 69 Relevant here is that contract-boundary-spanning governance comes in two primary forms. On the one hand, supply chains can be complemented with a dedicated governance contract that directly connects the two ends of a supply chain in addition to being indirectly connected via the underlying chain of contracts. Here, the Accord serves as a possible example. On the other hand, there are governance mechanisms that try to "blend in" in indirect chains of contractual relationships. Here, the Alliance provides an example as it complements its members' supply chains without explicitly interfering with the overall contractual structure of the chain.
These two basic approaches, one utilizing a dedicated governance contract and the other "blending in" into an existing contractual structure, can have very different outcomes from a legal perspective. The former, while clearly establishing a contractual relationship, may entail 67. Id. at 172-73. 68. Salminen, supra note 7. 69. While acknowledging the generally broad nature of the categorization, I followed Gereffi and colleagues in using the terms market governance, in which no mechanisms beyond market price are used by buyers to extend control beyond first tier suppliers; modular governance, where standards such as codes of conduct are cascaded to all members of the supply chain; and relational governance, where a broad range of contractually embedded techniques of communication and cooperation are used to create a level of control beyond privity. Id.
[Vol. 66 questions of conflict of contracts. 70 The latter is clouded by uncertainty in that it entails questions of whether the governance relationship can have legal effect in the first place and, if so, whether these effects are categorized as contractual, tortious, or something else. 71 Until at least recently, this uncertainty has been effectively used to limit liability for the inadequate governance of supply chains. 72 On the other hand, if liability is found in the case of the latter approach, then the former approach, a dedicated governance contract, could potentially be used to limit this liability. 73 To provide background to the discussion of the potential of these two basic modes of governance through contract in limiting liability and how law might respond, I will turn to the history of product liability law.

B. The Interaction of Governance Through Contract and Governance of Contract in the Development of Product Liability Law
Reflection on the relationship between governance through contract and governance of contract has a lengthy history in legal scholarship. 74 For the sake of argument, the development of product liability law could also be framed from this perspective. In particular, the historical evolution of product liability law provides a comparative example of how governance of contract has developed in different jurisdictions in reply to the challenges posed by new forms of governance through contract, in particular the widespread use of contractually organized distribution chains and disclaimers of liability. 75 This narrative can also provide a legal-doctrinal frame of reference for discussing the role of contract governance in supply-chain liability. In particular, supply-chain liability could be seen as a form of production liability, thus enabling the comparison of the development of contract governance in two arguably similar contexts. By the mid-nineteenth century, the mutual relationship of contract and tort under the common law was unclear. 77 For example, in the 1842 English ruling in Winterbottom v. Wright, the existence of a contract seemed to prohibit actions under tort against the parties to the contract even from those not in privity with them. 78 There are different explanations for this ruling, such as the unavailability in the circumstances of the case of an applicable tort action under early nineteenth-century English law. 79 Nonetheless, the ruling gave rise to the "privity" or "contract" fallacy that was later on used to outright bar actions under tort in cases related to contract. 80 This gave contracts a broad protective effect against claims under tort from nonprivy actors.
The privity fallacy was countered by the development of a general tort of negligence by American and English courts, in particular in relation to defective goods. 81 In the United States the 1916 ruling in MacPherson v. Buick 82 and in England the 1932 ruling in Donoghue v. Stevenson 83 are seen to establish the modern tort of negligence, allowing claims under negligence to coexist alongside contractual arrangements. 84 From the perspective of product liability, this development was driven by the rise of fragmented distribution chains where buyers were often reliant on a manufacturer's marketing to differentiate between products bought from retailers that were not easily appraisable, such as prepackaged foodstuffs or automobiles, and where mere users who had no relation to the chain of sales might have no remedy whatsoever under contract. Had the privity fallacy continued, it would have allowed manufacturers immunity from tort liability for defective goods. 85 Globalisation: The Great Unbundling(s) (Sept. 20, 2006), http://appli8.hec.fr/map/files/ globalisationthegreatunbundling(s).pdf. Baldwin's line of argument could be used to draw a comparison between product liability, one of the root causes of which was the fragmentation of distribution, and production liability, entailed by the fragmentation of production if one argues that production liability thus shares a fundamental root cause with product liability. There are, of course, also undeniable differences between product liability and any conceptualization of production liability, for which see, e.g., van  This, of course, is not the whole story. The common law traditions of England and the United States had begun to fracture more generally in their conceptualizations of contract and tort. This is visible, for example, in relation to contractual third-party beneficiary doctrines. In continental legal systems, third-party beneficiary doctrines were explicitly allowed for example by the French Civil Code and the German Bürgerliches Gesetzbuch. 86 In England, however, the doctrine was expressly prohibited from the mid-nineteenth century until the passing of the 1999 Contracts (Rights of Third Parties) Act. 87 Even after the passing of the Act, English law prefers actions under tort in some cases that would fall under contract in other legal systems. 88 In the United States, various contractual actions based on thirdparty beneficiary doctrines were increasingly allowed from the end of the nineteenth century onwards. 89 More interestingly, however, to counter some of the deficiencies of tortious actions in relation to liability for defective products, such as the need to show a defendant's negligence, American courts started experimenting with contractual actions that went much further than doctrines of third-party beneficiaries. 90 Actions perceived as contractual warranties allowed even non-privy actors, such as users of defective goods, to sue manufacturers or distributors under contract.

2018]
automobile under a contractual warranty theory while at the same time overriding contractual limitation clauses. 91 Such use of contractual causes of action, overriding both a lack of privity and contractual disclaimers, was not conceivable under English law. At the same time, however, contracts continued to cause problems in England for claims under both contract and tort. The rule of caveat emptor in contracts of sales of goods had been abolished through the 1893 Sale of Goods Act which gave rise to implied warranties of merchantability. 92 The law did not, however, limit the ruling out of this statutory liability. 93 As a result, manufacturers and sellers started providing "guarantees" which ostensibly provided a level of protection to users but, in practice, reduced the possibility and scope of claims under both implied warranties and the tort of negligence, even for personal injury. 94 The use of such guarantees spread to such an extent that legislative action finally prohibited the use of unreasonable contract terms, including "guarantees," for limiting liability through the 1977 Unfair Contract Terms Act. 95 In the United States, the use of actions founded in contract for overriding privity had its own opposition. 96 Product liability law, while retaining a stricter nature than normal negligence, was soon subjugated primarily to the field of tort, for example on the grounds that product liability was a policy concern and thus more appropriately governed by tort rather than contract. 97 As a result, liability for defective products was also capped because relegating product liability to the domain of tort meant that the economic loss rule, which typically rules out liability for pure economic loss under tort, became applicable. 98 While in theory pure economic loss may be recoverable under the tort of negligence in both England and in the United States, in practice recovery of such is restricted in both countries. 99  [Vol. 66 must be fulfilled but also the additional requirements for recovery of pure economic loss. 100 Doctrinal developments establishing liability for defective products took place also in other jurisdictions, such as Germany and France. Here, however, the initial parameters of contract and tort were radically different when compared to nineteenth-century common law. In particular, both French and German law were open to the use of contractual actions beyond privity in a broader scope than English or American law. 101 Under French law, contractual causes of action have played a key role in establishing product liability beyond privity for those connected to the chain of sales. Under so-called actions directes, a buyer could sue any other member of the chain of sales for breach of warranty. 102 Furthermore, French courts established that professional sellers are under an irrebuttable presumption of awareness of hidden defects in goods and thus cannot exempt themselves from damages arising out of such. 103 Those outside the chain of sales, such as mere users of products, had to rely on the French law of delict. Here, too, the courts developed a rule under which the putting into circulation of a defective good by a professional constituted negligence. 104 At the same time, the French law of delict allowed recovery of all types of damages without interference from contractual arrangements, so that this twoway approach allowed both those within the chain of sales and mere users similar possibilities for recovery of damages. 105 German law has similarly been more open to actions based in contract. In particular, contracts can more easily have an effect beyond privity through doctrines such as "protective effects of contracts towards third parties" or "transferred loss," which in many cases lead to pure economic loss being recoverable under contract in cases which under English law would require recourse to tort. 106 German scholarly discussion also supported the use of contractual techniques for establishing a general approach to product liability. 107 Nonetheless, the German Supreme Court adopted a delictual approach. 108

2018]
so, it developed a "duty to organize one's business" to overcome the challenges posed by the German law of delict, such as a lack of true vicarious liability. 109 This duty requires businesses to ensure that no third parties are injured during the course of business, with the added result that defects in products are used to imply a presumption of negligence. 110

C. The Accord and the Alliance: Two Polar Opposites of Governance Through Contract
The Accord and the Alliance represent, from a legal-structural perspective, two very different modes of supply-chain governance. The Alliance is based on a nonbinding model of supply-chain governance that blends into the supply chain's contractual structure. Through the Alliance, buyers coordinate their de facto possibilities for controlling working conditions in supplier factories. This control, however, does not unambiguously translate into a legally relevant connection upon which liability for inadequate governance could be grounded. This is because of the liability-limiting potential of the chain of contracts constituting the supply chain, where the nature of the indirect relationship between the buyer and its supplier's employees is clouded in uncertainty. On the other hand, as shown by the example of product liability, private law has developed specific mechanisms to counter the fragmentation of chains of contracts. Depending on the details of governance mechanisms and the parameters of governance of contract in relevant legal systems, liability for inadequate supply-chain governance could conceivably be found under a number of different theories of contract or tort. I will discuss some of these approaches in Part III.
The Accord, on the other hand, deviates from the traditional nonbinding model of supply-chain governance. It establishes a dedicated governance contract between buyers and their suppliers' employees, resulting unambiguously in a legally relevant relationship. Thus, there is little question of the contractual nature of liability between buyers and the representatives of their suppliers' employees for any promises made in mechanisms like the Accord. On the other hand, as seen in the example of product liability, contracts can be used to limit new forms of tort liability not only through their mere existence but also through more proactive means, such as in the form of "guarantees." Suppose that the risk of a buyer facing liability for inadequate governance of its supply chain under the theories discussed in Part III starts to outweigh the liability-limiting potential of noncommittal models of supply-chain governance. This raises the question of to what extent mechanisms such as the Accord can serve to limit any new [Vol. 66 liabilities opened up by developments in law. I will discuss the potential of using mechanisms such as the Accord in limiting supply-chain liability in Part IV.

III. EStablIShIng productIon lIabIlIty through prIvatE laW
A major strand of recent literature has focused on foreign direct liability, or whether a parent company can be held liable for the acts of its subsidiaries. 111 Less focus has been on the question of whether a buyer can be held liable for the inadequate governance of its supply chain. 112 The history of product liability shows that contract and tort causes of action can and have been used to establish liability in chains of contracts. 113 Here, I focus on the question of whether there is a tangible risk of liability under private law for a buyer's inadequate governance of its supply chain when using nonbinding governance mechanisms such as the Alliance.
More specifically, I look at the possibilities for actors lower down a chain of contracts, such as supplier employees, to sue buyers for inadequate supply-chain governance resulting in for example unfair or unsafe working conditions. This approach focuses on contract or tort duties arising out of the governance arrangements of buyers. With a few exceptions, the discussion here is focused on North American and European jurisdictions even if the substantive law in tort actions is in many cases that of where the damage occurs, for example the law

2018]
of Bangladesh or Pakistan as in recent high-profile cases of supplychain liability in Canada, the United States, and Germany. 114 This focus on certain North American and European jurisdictions is motivated in part by practical limitations and in part by my focus on supply-chain governance mechanisms that are embedded in contract. If the action lies in contract, then the substantive law applicable to the relevant contractual arrangements would most probably apply and, from the perspective of first-world buyers, this would in many cases probably be a North American or European jurisdiction familiar to the buyer. 115 In particular, in relation to governance contracts such as the Accord, choice of law becomes primarily contractual. 116 Another reason for focusing on these jurisdictions is that despite choice of law under tort pointing towards the law of the place where damage occurred, in many of the recent cases it has been argued that because of the common law tradition of, for example, Nigerian, Bangladeshi, and Pakistani law, recent developments in English common law may be relevant. 117 Of course, for the sake of argument my discussion here omits many of the numerous procedural complications of transnational litigation. 118 One case through which the legal nature of a buyer's supply-chain governance mechanism has been explored is that of Doe v. Wal-Mart   120 While the claim covered actions under a number of legal theories, including contractual and tort third-party beneficiary theories, 121 the crux of the complaint was that Wal-Mart had failed to enforce its code of conduct, which required suppliers to adhere to local laws and industry standards. Wal-Mart's alleged failure resulted in problems for supplier employees, ranging from excessive work hours and denial of pay or other benefits to lack of safety-related equipment, discrimination, and physical abuse. The courts rejected the claims. For them, the result hinged on the interpretation of the language used in Wal-Mart's then-applicable code of conduct. For example, in relation to the contractual third-party beneficiary theory, both the federal district and appeals courts found that the language used by Wal-Mart did not impose on Wal-Mart a duty to monitor suppliers but merely reserved the right for Wal-Mart to do so. 122 On the other hand, what the appeals court saw as a lack of significant control (in relation to negligent retention of control and supervision) and the lack of any undertaking to protect the employees (in relation to negligent undertaking) were key reasons for it finding that there was no tort duty of Wal-Mart to protect supplier employees. The result was dismissal for failure to state a claim.
I do not have room to discuss the case in more detail here. 123 In general, however, the treatment of the case is illustrative of two challenges faced by an approach focusing on the buyer's role in governing its supply chain in the United States. First, U.S. courts may be apt to interpret contractual language, such as a code of conduct included in a contract between a buyer and its supplier, restrictively unless there is a clear indication that supplier employees were intended beneficiaries 119 120. Earlier stages of the case also included claims filed by employees of Wal-Mart's competitors in Southern California.
121. By the time the complaint reached the appeals court, the four legal theories in focus were the contractual third-party beneficiary theory, joint employership, negligence (including third-party beneficiary negligence, negligent retention of control, negligent undertaking, and common law negligence), and unjust enrichment through the suffering of supplier employees.
122. The code of conduct contained wording such as "Wal-Mart will undertake affirmative measures, such as on-site inspection of production facilities, to implement and monitor said standards." See Doe, 572 F.3d at 680. But because this wording was found in a paragraph entitled "Right of Inspection" and because Wal-Mart would not experience any adverse effects if it did not monitor suppliers, the courts found that there was no promise on behalf of Wal-Mart to do so. 123

2018]
of the contract. Second, if the buyer is not directly controlling its supplier's employees, U.S. courts may have a hard time finding a duty under tort that would place liability on the buyer. This is due in part to the general tendency of the common law to emphasize that mere passersby are under no duty to help others. 124 Despite these challenges, some authors are positive that supplychain liability poses a tangible risk to defendants. 125 On the one hand, this is reflected in new cases of supply-chain liability. 126 On the other hand, a number of cases related to arguably similar forms of production liability, such as foreign direct liability and the subcontracting of environmental waste processing, have resulted in settlements or partial successes in courts. 127  [Vol. 66 First, the chances of success of supply-chain liability claims may depend on jurisdiction. 128 Some U.S. jurisdictions may be more accepting of supply-chain liability than others under any of the multiple possibly applicable theories of liability. 129 With regard to claims of negligence under English law and related common law legal systems, the reasoning of the 2012 ruling in Chandler v. Cape plc, 130 a case establishing a parent company's tortious duty of care towards its subsidiaries' employees under certain conditions, could provide a duty of care that is also applicable to supply-chain contexts in a broad array of common law jurisdictions. 131 When compared to English common law, the relatively lax requirements of contract formation and third-party beneficiary doctrines in jurisdictions such as Germany might make for easier claims under contractual theories. 132 At least in some civil law jurisdictions, such as French, German, and Dutch law, conceptualizations of delict focus more on claimants' rights rather than the common law focus on defendants' duties of care and thus may be more effective in obligating buyers to protect the physical integrity of supply-chain employees. 133 More generally, it seems that litigation related to the obligations of buyers towards their foreign supply chains or subsidiaries is only recently picking up in Europe. 134

2018]
may also prove important in relation to establishing duties of care or evidencing their breach. 135 Second, as seen in Doe v. Wal-Mart, much depends on the exact wording and nature of the obligations undertaken in codes of conduct and other governance mechanisms on the one hand and, on the other, on the de facto actions that buyers undertake to govern their supply chains. 136 Following this, one potentially crucial factor for supply-chain liability is the continuous development of supply-chain governance. In global supply chains buyers are gradually taking over the role of employers, governments, and unions. 137 Following, for example, Locke's description of the different phases of private governance, it seems that traditional "second stage" codes of conduct that have been interpreted as providing little or no direct rights to workers in the Doe v. Wal-Mart case are in many cases replaced with more elaborate models of supply-chain governance. 138 These newer forms of supply-chain governance, such as the Alliance, directly engage buyers in protecting supplier employees despite a lack of formal privity. Such direct action could stand a better chance of fulfilling the criteria for contractual or tort liability than the mere words included in earlier codes of Meeran, supra note 76, at 25-41; Enneking, supra note 111; Goldhaber, supra note 111; Revak, supra note 119. As noted, most of these cases concern foreign direct liability. Nonetheless they may provide legal developments that could apply in supply-chain contexts (see e.g., supra text accompanying notes 126 and 129-30). Finally, the German KiK case, discussed supra note 114, is a clean example of supply-chain liability. 136. For example, the codes of conduct used for governing supply chains come in many varieties that, based on their language and other characteristics, can be interpreted as more or less legally binding. See vytopIl, supra note 133. Similarly, in different jurisdictions some codes of conduct of Wal-Mart have been found more binding than others. See e.g., Revak, supra note 119.
137. For example, Phillips & Lim, supra note 112, at 350, write that [b]uyers are setting, monitoring, and enforcing standards in areas traditionally regulated by employers, governments, and unions, namely wages, hours, safety, discipline, and minimum age. They have recently moved into environmental regulation. Buyers send inspectors, punish violations, and teach, train, and coach their partners and allies. Buyers directly communicate with workers through hotlines and interviews, and sometimes state that they want to empower factory employees. Not surprisingly, workers seem to view buyers as allies, and this paternalistic role is largely recognized by unions and local management. In a similar vein, Anner, Bair & Blasi, supra note 2, at 30, discuss the resemblance of a mechanism such as the Accord to jobber's agreements between unions, contractors, and employers in twentieth-century United States.
[Vol. 66 conduct. 139 Ultimately, a more profound understanding of the different ways in which buyers exert de facto control over their supply chains by various mechanisms of governance could help concretize general normative requirements of adequate supply-chain governance. 140 Finally, one may also reflect on historical developments in other fields, such as that of the development of product liability law. Currently, no law of production liability exists on terms similar to product liability law. Nonetheless, the parallels in the historical development of product liability law and current developments in relation to production liability have not gone unnoticed. 141 The rise and gradual polishing of a multitude of different legal doctrines, ultimately resulting in a distinct form of relatively strict tort liability for defective products, were responses to perceived injustices brought about by fragmented distribution chains. While it seems unlikely that buyers will any time soon face a similar liability over their supply chains as manufacturers face in relation to harm caused by defective products, there is momentum towards establishing some level of production liability. 142 In particular, this momentum towards production liability is reflected in the number of recent settlements and partial successes in court. 143 This momentum could affect risk-averse buyers in the sense that they will need to consider how they can minimize the potential risks of production liability entailed by the use of increasingly advanced governance mechanisms such as the Alliance.
Iv. a ShIFt oF paradIgm In Supply-chaIn govErnancE: dIrEct govErnancE contractS aS vEhIclES For controllIng lIabIlIty?

A. Direct Governance Contracts as a Remedy for the Uncertainties of Earlier Governance Mechanisms
As seen in Part II, private ordering through contract is a well-established way of controlling liability. The inherent possibilities of contracts for limiting liability are what makes the Accord, as a governance 139. For just a few examples, this could take place under so-called Good Samaritan tort duties or even by pushing a court to find that, despite vague language, the scope of a contractual obligation was de facto intended to benefit or protect third parties. See, e.g., Phillips  , argue that there is a cautious but significant move in court practice away from the dogma of strict separation of responsibilities between parent companies and subsidiaries towards a tort law based focus on liability based on control. This is an approach that is not limited to corporate structures but that can also be applied to supply chains. Going further, for example in Das v. George Weston the trial court found that several policy factors supported a more general duty of care on buyers towards supplier employees, even if these factors were outweighed by other policy factors. Das v. George Weston, Ltd., 2017 ONSC 4129, ¶ ¶ 451-52.

2018]
contract directly connecting buyers to representatives of their supplier employees, particularly interesting from the perspective of supplychain governance. Traditional CSR mechanisms are intended to be nonbinding from a legal perspective. It is uncertain whether they can give rise to liability. Nonetheless, as seen in Part III, such liability has been discussed and tested in courts to the extent that there is a tangible risk of liability for inadequate supply-chain governance. From the perspective of the traditional, nonbinding CSR paradigm, buyers seem to have two choices for controlling liability. On the one hand they can implement governance mechanisms that are in practice effective in governing their supply chains, for example by actively working to provide safe working conditions for supplier employees. This leads to a risk of liability in case the active governance efforts prove inadequate. On the other hand, they may try to avoid effective governance altogether. This risks not only repercussions from consumers and regulators but there is also a risk of liability for inadequate supply-chain governance.
Here, the model of the governance contract represented by the Accord steps in. In establishing a direct governance contract between buyers and supplier employee representatives, the Accord not only establishes what at least appears to be an effective mechanism for the governance of working conditions at supplier facilities but, at the same time, provides what in the supply-chain governance context is a new kind of vehicle for controlling the risk of liability. To highlight the paradigm-changing potential of direct governance contracts in balancing active governance and control of liability, I will next use the Accord and its provisions as an example of the potential scope of controlling liability inherent in such mechanisms.

The Substantive Reach of Governance Contracts
It is up to the parties to define how broad or narrow the substantive scope of their contract is. Here, one of the key critiques raised against the Accord is its narrow substantive scope. 144 The Accord is restricted to improving fire and building safety in the signatories' supplier factories and negating the harmful effects that undertaking these improvements may have on factory workers. A large number of issues are thus beyond the scope of the Accord. To give some examples, the Accord does not discuss the fairness of wages or benefits in general, instead limiting this discussion to what could be a fair amount of compensation for unemployment due to factory downtime caused by a buyer requiring specific repairs. 145 Nor does it discuss broader human [Vol. 66 rights issues, such as combating discrimination, except for in the sense that the effects of factory downtime caused by buyers' requirements for specific repairs must be applied in a nondiscriminatory manner. 146 Ultimately, the scope of any agreement depends on the interpretation of its wording. In the case of the Accord, it is difficult to see its scope interpreted as covering issues more broadly than what is explicitly stated as its narrow scope. On the other hand, it is easy to imagine that the scope of a similar agreement could be much broader, covering for example a whole code of conduct. 147 In such a case, a buyer might attempt to use a governance contract to control its liability over its supply chain in relation to a broad range of issues ranging from personal injury to social and environmental issues.
In addition to the parties' agreement, legal systems limit the extent to which contractual arrangements can be used to govern the relationship of the parties. Issues considered important for public policy are typically excluded from the sphere of private ordering. For example, in the United States, liability for negligence can generally be excluded through contractual arrangements. 148 Interests protected by public policy, however, are exempt from this general rule and thus for example liability for intentionally or recklessly caused harm cannot be exempted, employers cannot limit their liability in negligence towards employees, and common carriers or public utilities cannot limit their liability in negligence towards customers. 149 Similarly, liability for negligence cannot be excluded through contractual arrangements deemed as unconscionable. 150 What is considered public policy depends on the underlying legal system. 151 For example, national laws differ greatly in their stance on the acceptability of arbitration for resolving certain kinds of disputes. 152 While in the United States consumer arbitration is generally accepted, the public policy of the European Union classifies pre-dispute arbitration clauses between businesses and consumers as categorically unenforceable. 153

2018]
Exceptions to the freedom of contract based on public policy could have an effect on governance contracts. Omri Ben-Shahar and James White point out in their study of American automotive supply chains, that the contracts they encountered between buyers and first-tier suppliers could in many cases be seen as unconscionable. 155 Even more scrutiny might be placed on a contract between clearly asymmetrically powerful parties, such as first-world buyers and the employees of third-world suppliers, even when the latter are represented by unions, as in the case of the Accord. A direct governance contract such as the Accord might also liken supplier employees to the buyer's own employees so that both would fall under public-policy-related employee protection schemes. 156 On the other hand, as discussed below, contractual arrangements can also be used to remove themselves from the ambit of specific public policies. 157

The Personal Scope of Governance Contracts
A traditional contract law perspective emphasizes that a contract only binds its parties. Thus even if a governance contract can have practical effects on an indeterminable class of actors, it cannot legally bind such a class ex ante. An example is the Swedish case Arica Victims KB v. Boliden Mineral AB. 158 That case concerned the sale in the mid-1980s of toxic sludge from a Swedish mining company to a Chilean company that promised to appropriately handle the sludge. Soon after, the Chilean company went into liquidation and dumped the sludge near the desert town of Arica. Desert winds blew arsenic from the sludge mounds into town, resulting in an allegation of hundreds of cases of arsenic-related illnesses in local inhabitants. While related to a specific contract, such cases cannot easily be controlled through ex ante contractual governance mechanisms.
Determinable classes of actors, such as suppliers and supplier employees, can be bound under a governance contract. A further question concerns the mechanism of adhesion. From a practical perspective, supplier companies can relatively easily be managed as signatories to an agreement. Supplier employees, on the other hand, may constitute such a large and changing mass of actors that it is practically impossible to maintain a direct contractual relationship between each and every employee. Here, union representation provides an alternative. In the case of the Accord, for example, global and local unions, as It can also be asked to what extent potential parties can agree on entering a governance contract. In this sense, the Accord is an extremely successful example. It is perceived as improving the situation of Bangladeshi garment workers and thus has garnered enough goodwill to be signed by global and local unions representing a major part of supplier employees. 159 In addition to or instead of strategies emphasizing the fairness of the deal, however, buyers might either explicitly or unwittingly end up using their economic clout or suppliers' or supplier employees' distress to coerce them into accepting a less than ideal deal. 160 Another issue concerns the potential of third parties to enforce and be bound by a governance contract. Third-party beneficiary doctrines provide one possibility for doing this. 161 Under such doctrines actors who are not parties to a governance contract in a strict sense can enforce the contract if they are seen as its intended beneficiaries. 162 However, if actors choose to rely on their status as third-party beneficiaries of a contract, they are typically also bound by any limitations of liability that the contract entails. 163 In the case of the Accord, characterizing, for example, nonunionized supplier employees as third-party beneficiaries is conceivable as the intention behind the Accord is specifically to improve the working conditions of the employees of the signatories' suppliers. Thus, workers employed at a supplier factory that is covered by the Accord may be able to draw upon the benefits of the Accord (and in doing so be confined by its limitations of liability) even if they are not members of one of the Accord's union signatories.
Finally, through mechanisms such as the Accord, buyers may have considerable de facto power in legitimizing standards of compensation that have been mutually agreed by buyers and unions and therefore perceived as generally fair. 164

2018]
harmed supplier employees to voluntarily accept such standards on a third-party beneficiary basis or via settlement agreements, especially as the alternative, litigation, may see courts and tribunals similarly looking for guidance in the standards set by the Accord. Furthermore, litigation can generally be a harrowing experience for supplier employees who, even if represented pro bono and with support from unions, may not have the means to sustain themselves or remedy their injuries during the duration of litigation. 165 Thus mechanisms such as the Accord have the potential to control liability even beyond those who are legally bound by them.

C. The Accord as an Example Framework for the Contractual
Control of Liability

Controlling Liability Through Contractual Guarantees
As discussed in Part II, under English common law manufacturers used "guarantees" to limit liability for defective products under tort. While the guarantees explicitly provided for a certain degree of liability, at the same time they capped this liability from that which could have been possible under English sales law (e.g., under the implied warranties of merchantability first implemented in the 1893 Sales Act) and tort law (e.g., with regard to liability for personal injury or death). Such guarantees became standard practice to the extent that the 1977 Unfair Contract Terms Act was passed in part to prohibit them. This example is meant to be illustrative of the point that not only explicit limitations of liability can be used, but also that what appear to be explicit guarantees of liability can be used to the same extent.
Take the example of the Accord, which is intended to improve the position of supplier employees in Bangladesh. In addition to generally organizing the identification and renovation of hazardous factories, it makes promises aimed at supplier employees. For example, Article 13 of the Accord states that [s]ignatory companies shall require their supplier factories that are inspected under the Program to maintain workers' employment relationship and regular income during any period that a factory (or portion of a factory) is closed for renovations necessary to complete such Corrective Actions for a period of no longer than six months. . . [sic] Failure to do so may trigger a notice, warning and ultimately termination of the business relationship as described in paragraph 21. 166

[Vol. 66
This is a complex clause offering multiple possible interpretations. 167 One interpretation could see the clause as a contractual guarantee of a maximum of six months' wages to employees who are out of work due to factory renovations. Following this interpretation, the Accord could be seen as not only an acknowledgment of a limited level of liability but also as a substantive limitation of such liability, capping the damages owed by a buyer to each employee to the amount of the employee's six months' wages.
Other situations governed by the Accord are not as straightforward. For example, under Article 14 of the Accord buyer signatories promise to undertake "reasonable efforts to ensure that any workers whose employment is terminated as a result of any loss of orders at a factory are offered employment with safe suppliers." 168 On the one hand, here liability is limited more vaguely by subjecting it to the interpretation of "reasonable efforts." On the other hand, the apparent example of "reasonable efforts" provided by Article 14 of the Accord, "actively working with other suppliers to provide hiring preferences," seems to set the bar for fulfilling this limitation of liability quite low for buyer signatories.
As in the example of English "guarantees" that were later limited by legislative action, the possibility for using various limitations of liability depends on the parameters of the underlying legal system and the availability of legal safeguards. Similarly, a key challenge in using contracts to limit liability in supply chains is presented by public policy that may limit the use of limitations of liability, for example, in relation to weaker parties. As discussed next, however, contractual arrangements can be used to alter the framework of liability so that public policy embedded in specific legal systems might no longer govern a mechanism such as the Accord.

Contractual Arrangements for Displacing National Legal Safeguards
Contractual arrangements can be used to control access to remedies embedded in national legal systems. Disputes can be relegated 167. Instead of construing this provision as a guarantee of six months' wages to employees, it could be construed merely as buyers promising to attempt to incentivize their suppliers to do so through the supplier incentives mentioned in Article 21 of the Accord. Under Article 21, buyers are required to implement a notice and warning procedure, ultimately leading to termination of business against noncompliant suppliers. For example, following the ruling in Doe v. Wal-Mart, it could be argued that the Accord does not explicitly mention any adverse consequences to a buyer if it fails to enforce this provision and thus the provision creates no real obligation on behalf of the signatory. However, the Doe v. Wal-Mart case concerned the question of whether enforceable rights were granted to supplier employees under third-party beneficiary theories. Because the Accord is a direct contract between buyers and the representatives of supplier employees, it seems more reasonable that supplier employees or their representatives should be able to sue buyers for damages arising out of a failure to adequately incentivize suppliers to comply. Finally, Articles 22 and 23 of the Accord require buyers to provide not only warnings and threats of termination to suppliers but also financial incentives and a commitment to long-term sourcing relationships in Bangladesh. These additional commitments could support the case for buyers being liable for failing to adequately incentivize suppliers.

2018]
to dispute resolution procedures or legal systems that are more favorable to certain types of arguments or cases. Incorporating procedural uncertainties into agreements may also have an effect, for example if they are later on used to coerce settlement through the threat of extended litigation over issues such as jurisdiction or governing law. 169 Taken together, as seems to be the case under the Accord, these three factors support one another in creating a procedural framework for controlling supply-chain liability. First, relegating dispute resolution to specific mechanisms, such as arbitration, affects the legal framework in which contractual arrangements are interpreted. Arbitration can provide an effective alternative to state courts for resolving disputes between commercial parties. 170 In other contexts, such as in relation to claims by consumers, employees, and supplier employees, the fairness of arbitration has been disputed. 171 In some jurisdictions even domestic arbitration may allow actors to distance proceedings from legal safeguards embedded in national law that are intended to protect weaker parties such as consumers and employees. 172 International commercial arbitration in particular relies on the process drawn up by the parties and an international enforcement framework without substantive review that makes it difficult to overturn awards even on public policy grounds. 173 This framework allows parties to draft arbitration agreements that can override national procedural safeguards for the sake of party autonomy and certainty. More generally, international investment arbitration, relying to a large extent on the same framework as international commercial arbitration, has been criticized because its institutional structure is poorly equipped to perform public law adjudicatory functions. 174 [Vol. 66 As a practical example, Article 5 of the Accord contains an arbitration provision. Under the Accord, arbitrators would deal with what are essentially labor disputes between first-world companies and third-world supplier employees. The employees are represented by unions who have had their say in the drafting of the Accord and thus it is difficult to critique the arbitration provision of the Accord as inherently unfair. Nonetheless, the arbitration provision moves the Accord towards a legal space where the sharpest focus is on the text of the Accord while any safeguards embedded in potentially relevant national public policies are respectively blurred.
Second, the increased focus on the parties' agreement that follows from subjugating disputes to arbitration can be increased by choice of law. This can be done by choosing a legal system allowing parties greater freedom with respect to private ordering. Alternatively, parties may opt to generally distance their agreement from national legal systems into a more transnational sphere. Transnational law as such is a problematic concept. 175 Instead, it may suffice to rely on transnational distancing, with which I understand the extra deference that may be given to contractual arrangements that are simultaneously embedded in more than one legal system. For example, Article 7 of the United Nations Convention on Contracts for the International Sale of Goods (CISG) requires that in interpreting the convention "regard is to be had to its international character and to the need to promote uniformity in its application and the observance of good faith in international trade." 176 This is generally understood as a requirement to abstain from using material embedded solely in a national legal system in interpreting the CISG. 177 Similar approaches can be embedded ad hoc in individual contracts. 178 In practice, this distancing may move the interpretation of contracts away from specific national legal systems towards any "common ground" identifiable for example in the legislative history of international instruments or a comparison of different legal systems. 179 In such circumstances, concepts regarded as nearly universal, such as the binding nature of contracts, may be more

2018]
readily acceptable than concepts ingrained in national legal systems, such as local public policies on fair contracting.
Again, the Accord may provide a practical example. It has over 200 signatories from dozens of different jurisdictions. 180 No single law has been explicitly chosen to govern the procedural and substantive issues of disputes relating to the Accord. 181 Instead, a flexible procedure, based on the 1985 UNCITRAL Model Law on International Commercial Arbitration, with its 2006 amendments, has been adopted. 182 The Model Law incorporates mechanisms for identifying the procedural and substantive laws applicable in case of individual disputes. 183 If the parties to a dispute cannot agree on these issues, the choice is ultimately relegated to arbitrators to be made on the basis of the circumstances of each dispute. As noted, any dispute may involve parties to the Accord from a broad range of jurisdictions. This means that, in principle at least, in a dispute the legal effects of the Accord could end up being interpreted from the perspective of any of the different legal systems in which parties to the Accord are domiciled or otherwise connected to.
At the same time, the Accord contains no provision relating to its uniform interpretation such as that included in Article 7 CISG. Nonetheless, it is reasonable to think that such a provision could be implied into the Accord. Because of the broad array of jurisdictions potentially relevant for interpreting the Accord in individual disputes, contrasting approaches to private ordering adopted in different jurisdictions might lead to contradictory outcomes. The dangers of contradictory interpretations and attempts at forum shopping could give rise to a notion of uniform interpretation of the Accord. If so, arbitrators, already primarily focused on the parties' agreement, could be inclined to focus even more on accepted transnational principles, such as pacta sunt servanda, instead of considering public policy safeguards embedded in some of the potentially applicable national legal systems.
Third, and finally, any tendency to focus on the four corners of a contract, as supported by choice of procedure and choice of law, might be further strengthened by procedural uncertainties incorporated into the contract. Open-ended procedural frameworks may be necessary 180 [Vol. 66 in contractual arrangements that involve hundreds of potential parties and, as a result, entail different combinations of potentially relevant jurisdictions for each possible dispute. This is because in such cases it can be impossible for parties to agree on choice of law and forum beforehand. On the other hand, open procedural frameworks may also be used by disgruntled parties to contest and prolong proceedings and the enforcement of any judgment or award. They can thus become a de facto limitation of liability by making the resolution of disputes and the enforcement of their outcomes difficult and uncertain. Again, the Accord may provide an example. The Accord is not anchored in any specific jurisdiction or substantive law. Instead, the laws used to govern a dispute related to the Accord seem to be chosen ad hoc for each dispute. The mechanisms for doing this are provided by the Model Law, which provides for an open-ended procedure for defining both the procedural and substantive laws applicable to a dispute. With regard to procedural law, the Model Law equates it with the law of the place of arbitration. 184 In case the parties cannot reach an agreement over the place of arbitration and, consequently, the procedural law, Article 20 of the Model Law requires arbitrators to determine the place of arbitration by "having regard to the circumstances of the case, including the convenience of the parties." Similarly, if the parties cannot agree on substantive law, under Article 28 of the Model Law "the arbitral tribunal shall apply the law determined by the conflict of laws rules which it considers applicable." 185 Both procedures can involve several contingencies and complications. 186 Finally, the Model Law was originally designed to be enacted by a state as its national arbitration law governing international arbitration. 187 The use of the bare, unenacted Model Law provisions may also increase uncertainty because it leaves open a number of procedural issues that states are supposed to decide upon when enacting the Model Law. 188 Other procedural problems may also arise, as is clear from the first publicly available decision related to arbitration under the Accord [Vol. 66 in controlling their supply chains. With the adept use of liability disclaimers, this control could be seen as inexpensive in comparison to the liabilities that it opens up. The cost-benefits may be great in particular in relation to the earlier paradigm of exerting control without privity, which imposes heavy requirements for buyers to keep up the appearance of noninvolvement in their supply chains for example by using specific noncommittal language.
The paradigm represented by the Accord is not necessarily new. Similar structures may have been used earlier on in business contexts. In the specific context of CSR, however, it is definitely a new development in relation to the earlier paradigm, that of CSR which is nonbinding in the sense that it is both exclusive of third-world actors and unenforceable by them. The older, nonbinding paradigm could be updated by adding increasing layers of noncommittal language, but in the end it always relies on contractual boundaries for its liability mitigating effect. Problems associated with such an approach include losing any positive image effects on consumers and regulators as problems in supply chains persevere, the gradual compromising of their liability reducing capabilities, and the little control over supply chains that they allow because control tends to create privity in fact. Compared to the nonbinding paradigm, the new paradigm represented by the Accord, which is inclusive of third-world actors and allows them to enforce the agreement, overcomes this by replacing the earlier reliance on lack of privity by proactive private ordering. Binding, inclusive, and enforceable become the slogans of the new paradigm instead of nonbinding, exclusive, and unenforceable, the failed words of the old paradigm.
But is not some level of liability better than no liability? CSR mechanisms akin to the Accord can be seen as similar to contractual guarantees in that they provide a certain level of rights and remedies to supplier employees, but these rights are easily capped through contractual provisions. Especially if one takes the perspective that there would otherwise de facto be no such liability due to the economic and legal challenges of transnational litigation, then some level of acknowledged liability might be seen as a positive improvement. However, the question then becomes whether the level of liability acknowledged in such a mechanism can be seen as fair. This hinges on the relative power of the actors and, in the end, in any benevolence that firstworld companies are willing to show despite their quest for profit. As has been seen under the nonbinding paradigm, the end result can be a dismal failure. The same critique applies under the new paradigm, even if it is dressed in the clothes of an enforceable and inclusive contract. Thus, in the end, it is unclear to what extent the inclusive and enforceable paradigm has the power in practice to improve labor conditions more effectively than the earlier exclusive and unenforceable paradigm.
On the other hand, the new paradigm may give adjudicators a new possibility to intervene. This would require the use of doctrines embedded in national legal systems, such as fairness or unconscionability, to strike down liability limiting clauses contained in contracts such as the Accord. This seems at least from a legal-technical perspective easier than current attempts at constructing supply-chain liability between actors not in privity. Several problems, however, remain. Key among these is the transnationalization of governance contracts such as the Accord, which pushes them away from the applicable sphere of national regulatory safeguards. Another problem may be the increased deference shown by national adjudicators towards private ordering, in particular when combined with transnational disputes. Thus, if mechanisms such as the Accord proliferate, the role of courts and tribunals will be crucial in assessing the reasonable bounds of private ordering in transnational contexts. This results in a need to develop doctrines of contractual safeguards so that they are more clearly applicable and practicable also in the transnational sphere.