Corporate social responsibility in the global value chain: A bargaining perspective

Breaches of corporate social responsibility (CSR) in global value chains (GVCs) pose a managerial challenge for multinational enterprises (MNEs) and threaten both their reputations and global sustainability. While an MNE-centric perspective on these issues has dominated existing international business research, we show that a dynamic view of bargaining among actors in the GVC can yield novel insights. We draw on coalitional game theory and develop a model where an MNE collaborates, monitors, and negotiates prices with a supplier whose CSR breaches may be revealed by the MNE, external agents, or remain hidden. Our model illustrates how MNEs may face a hold-up problem when irresponsible actions by suppliers are made public, and the suppliers have the power to engage in opportunistic renegotiation. Interestingly, we show that greater monitoring by MNEs, if not combined with specific strategies, can have negative consequences by weakening the MNE’s bargaining position and, in some cases, even prompting more irresponsible actions by the suppliers. Our model advances international business research on GVC sustainability and has important implications for managers and researchers alike.


INTRODUCTION
Decades of globalization have triggered the rise of large, technologically advanced firms in emerging markets that supply developed country multinational enterprises (MNEs) with essential products and services (Dicken, 2007). By connecting with suppliers in diverse locations, MNEs can contribute to higher levels of value creation for all involved actors (Mudambi, 2008). At the same time, there is a growing recognition that social and environmental responsibilities are often neglected at supplier sites (Barrientos, Gereffi, & Rossi, 2011;Stringer, Hughes, Whittaker, Haworth, & Simmons, 2016). Notorious examples include human rights violations at Apple's Taiwanese supplier Foxconn (Lee, Mol & Mellahi, 2016), child labor at Rangan Export, an Indian rug supplier of the Swedish furniture retailer IKEA (Bartlett, Desain & Sjoman, 2006), and the use of Asian sweatshop suppliers in the 1990s by the American sportswear leader Nike (DeTienne & Lewis, 2005). More recently, the Spanish fast-fashion company Zara was the target of an international scandal after the owner of Bravo Tekstil, one of its largest Turkish suppliers, left hundreds of workers uncompensated for months' worth of labor (Fast Company, 2017).
Irresponsible behavior consists of actions that violate social and environmental expectations, as judged by a majority of impartial global stakeholders (Armstrong, 1977), and can thus be regarded as a breach of corporate social responsibility (CSR) policies imposed by MNEs (Narula, 2019). Such behavior at supplier sites is potentially deleterious for MNEs, who are often held accountable for their suppliers' social and environmental impact (Kim & Davis, 2016;Wettstein, Giuliani, Santangelo, & Stahl, 2019). Irresponsible supplier behavior may undermine an MNE's reputation, global brand equity, and future market prospects and business opportunities (Asmussen & Fosfuri, 2019;Maggioni, Santangelo, & Koymen-Ozer, 2019). For example, the workers at Bravo Tekstil started a campaign against Zara as a reaction to their missing wages, thus harming the social brand of the Spanish apparel retailer; and the 2012 riots at Foxconn wiped USD 30 billion off Apple's stock price in a single day (Seeking Alpha, 2012).
We explore the conditions under which these irresponsible behaviors emerge at supplier sites and examine how MNEs can reduce their occurrence. While IB research has made important contributions in explaining how the MNE uses its strategic leadership in global value chain (GVC) relationships to address irresponsible supplier behavior, we argue that the dominant perspective is overly ''MNE-centric'' (Hennart, 2009) and falls short on three accounts. First, this literature underestimates the power dynamics of client-supplier relationships (Denicolai, Strange & Zucchella, 2015), which challenge the view of a static bargaining power distribution that necessarily favors the leading MNE. For example, Foxconn possesses unique anodization and rapid assembly capabilities that make them strategically important to Apple (Kanematsu, 2017), thereby reducing Apple's bargaining power. Second, many suppliers operate in contexts with varying degrees of vigilance and scrutiny (Kim & Davis, 2016;Lund-Thomsen, 2020). These contexts can limit the suppliers' sensitivity to responsible conduct and increase the asymmetries in reputational damage in their relationships. For example, Zara's reputation was more vulnerable to its supplier's conduct than the supplier's own reputation was. Third, MNE-supplier relationships are often complex and defined by limited visibility (Kim & Davis, 2016;Narula, 2019). Suppliers in developing countries, like the Indian rug suppliers of IKEA, can therefore prioritize efficiency over sustainability in response to the pressure to reduce costs by their multinational clients.
To address these limitations, we draw on research that recognizes the dynamic role of bargaining power and the agency of suppliers in the sustainability upgrading of GVC activities across industries (De Marchi & Di Maria, 2019) and questions the efficacy of monitoring mechanisms (Jeppesen & Hansen, 2004;Locke, Qin, & Brause, 2007) to engage local suppliers who are often resourceconstrained and work in institutionally weak contexts (Goger, 2013;Golgeci, Makhmadshoev, & Demirbag, 2021;Lund-Thomsen, 2020). We develop a coalitional game-theoretic model of a GVC dyadic relationship to take these insights to IB research. In our model, an MNE and a (potentially irresponsibly behaving) supplier interact strategically and negotiate prices over time. The supplier may decide to engage in misconduct, which may be discovered either by outside agents or by the MNE, who in turn can choose to ignore it, reveal it to the outside world, work privately with the supplier to correct it, or terminate the relationship. Based on this, we study issues regarding bargaining power, asymmetries in reputational damage, and monitoring. In particular, we focus on a conflict of interest where suppliers are the wrongdoers while MNEs prefer sustainable practices (but, as discussed below, our model is sufficiently general to cover other cases).
We claim three main contributions to the literature. First, we depart from an MNE-centric perspective by abandoning the assumption that the of the MNE is all-powerful and by considering the supplier's agency and bargaining power. Allowing for the possibility of asymmetries in reputational vulnerability between MNEs and suppliers enables us to demonstrate that MNEs may be subject to opportunistic renegotiation by suppliers and suffer a hold-up problem as irresponsible actions in their GVCs are revealed. While power within GVCs has been discussed in the IB literature, we extend this work by emphasizing the changes in bargaining position driven by the reputational damage diffused to the MNE by the supplier's irresponsible behavior. Second, we explore the consequences of monitoring for the MNE, showing that in some cases, having greater information on a supplier's irresponsible conduct can exacerbate this problem and that the expectation of monitoring can even influence supplier incentives towards more irresponsible action. We believe these counterintuitive insights are novel to the IB literature. While existing research has reported that monitoring can be ineffective (Lund-Thomsen & Lindgreen, 2014), we show that it might even have unintended negative consequences that need to be managed. Third, we identify strategies MNEs can pursue to avoid and remedy these pitfalls, including credible commitments to terminate a relationship with an irresponsible supplier or publicly announce any irresponsible action discovered through monitoring. While seemingly limiting the actions available to MNE managers, we show that these strategies can have indirect, strategic effects that ultimately improve MNE performance.

GVC and Irresponsible Supplier Behavior
The fragmentation of production and the international dispersion of related tasks and activities have favored the emergence of cross-border sequential chains referred to as GVCs. GVCs describe the ''full range of activities that firms […] perform to bring a product from its conception to end use and beyond […] in inter-firm networks on a global scale'' (Fernandez-Stark & Gereffi, 2019: 55). GVCs have been regarded as mechanisms that ease developing country firms' economic and social catch-up (Mudambi, 2008). Firms in developing countries participating in GVCs can move to higher valuecreating activities in production and improve their technology, knowledge, and skills (Gereffi, Humphrey, & Sturgeon, 2005). Their participation can also improve the overall well-being of workers and local communities by enhancing working conditions, labor rights protection, and environmentally respectful production (Gereffi & Lee, 2016).
As the cases of Apple, Ikea, Nike, and Zara indicate, however, GVC economic upgrading often fails to translate into the social upgrading of workers and communities in host countries (Barrientos et al., 2011). For example, the case of Foxconn shows how a contract manufacturer can supply global brands such as Apple and Nokia and, thereby, provide employment to millions of workers. At the same time, Foxconn was accused of unfair working conditions, including involuntary and often unpaid overtime and inadequate safety measures. Similar types of irresponsible GVC behavior have been documented for Nike, which was in the spotlight for the irresponsible labor practices at its sweatshop contractor companies in Indonesia, Vietnam, and Pakistan in the 1990s. The garment industry was again under scrutiny after the Rana Plaza factory collapse in 2013, whereby more than 1100 workers lost their lives while supplying garments to major Western fashion brands.
Irresponsible practices at the facilities of suppliers recur despite the effort of MNEs to promote CSR policies in their GVC relationships. The success of such CSR initiatives is often limited by the inherent tensions between MNEs and their lead suppliers (Lund-Thomsen & Lindgreen, 2014), compounded by the competition that suppliers face to respond to the pressure for cost reductions and the weak institutional conditions in which they operate (Goger, 2013;Golgeci et al., 2021;Lund-Thomsen, 2020).

GVCs in IB Research
In reflecting the origins of the field, IB research on GVCs has traditionally held the MNE as its logical unit of analysis. For example, research has explored the role of the MNE as the orchestrating firm (Kano, 2018;Strange & Humphrey, 2019), how MNEs influence engagement in GVCs across countries and industries (Fortanier, Miao, Kolk, & Pisani, 2020), and their contribution to the creation and evolution of local hubs of capabilities in specific activities of the GVC (Gereffi & Lee, 2016). In this stream, advanced country MNEs typically coordinate GVCs, with cross-border trade of production inputs and outputs taking place within their networks of arm's-length suppliers. More recently, a relational perspective on GVCs has gained prominence. Here, the MNE is seen to leverage its firmspecific advantage to manage the bounded rationality and reliability of the partners involved (Asmussen, Chi, & Narula, 2022;Kano, 2018) and create an organizational context supporting the multiple stages of value creation within the GVC network. Thus, through its strategic leadership, the MNE aligns the interests of the different parties by reducing information complexity, creating incentives to increase the parties' effort to make good on open-ended promises, and supporting value creation along the different stages of the value chain.
Existing research has focused on governance as central to achieving these goals, with the MNE taking control of decisions that define the mix of internalized and externalized activities (Buckley, 2014). It also identifies the most advantageous geographical configurations of the GVC by examining the nature of industrial clusters, labor cost differentials, and specialization across locations (Asmussen, Pedersen, & Petersen, 2007). Finally, the MNE orchestrates the relationships within the GVC network to facilitate the performance goals of the individual participants and the sustainability of the whole value chain (Sturgeon, Van Biesebroeck & Gereffi, 2008). Large and highly visible MNEs are incentivized to protect their ''social brands'' -expressed in sustainability reports, corporate communications, advertising campaigns, and trademarks -from reputational damage (Asmussen & Fosfuri, 2019). As these brands may be undermined by social and environmental disasters at supplier sites, lead firms often strive to ensure CSR compliance and sustainability throughout the GVC. Castaldi, Wilhelm, Beugelsdijk and van der Vaart (2023) suggest that they can do so with a 'stick-and-carrot' approach (using audit-and cooperation-based governance), but more effectively so in strong institutional environments.
Notwithstanding these important contributions, our understanding of the dynamics of irresponsible supplier behavior within the GVC remains incomplete. In particular, an MNE-centered perspective is subject to at least three limitations: a static understanding of power relationships in GVCs, the overlooked CSR perspective of suppliers, and the allegedly unequivocal beneficial role of monitoring. We discuss and juxtapose each of them with recent advancements in operations management, political economy, and business ethics research below.

Power relationships in GVCs
In IB research, the MNE's CSR practices are seen to be vertically extended to its global suppliers, while supplier local environmental conditions and practices have been mostly overlooked (Lund-Thomsen, 2020). The MNE is thus expected to ensure compliant behavior by investing in relational capital to facilitate coordination and reduce the hazards of imperfect effort by partners (Kano, 2018). However, power dynamics between the MNE and its suppliers are subject to change. The rise of suppliers with strong capabilities of their own unleashes complex dynamics in the GVC relationships, which can undermine the power of the MNE and increase its dependence upon its suppliers (Denicolai et al., 2015). Such a dependence exposes the MNE to reputational vulnerability with potentially deleterious effects on its social branding. We have an inadequate understanding of the bargaining situations between MNEs and their suppliers when the traditional power asymmetries change.

Supplier perspective on CSR in GVCs
The MNE-centric view risks overlooking the supplier perspective on CSR in GVC relationships. The scrutiny to which MNEs are exposed worldwide motivates their initiatives to impose CSR compliance on their suppliers in developing countries (Amaeshi, Osuji, & Nnodim, 2008). However, suppliers in these institutional contexts (in the cases mentioned above, companies such as Rangan Export and Bravo Tekstil) are less likely to be held accountable for CSR violations as they face limited stakeholder pressures. Consequently, their actions are often motivated by factors other than accountability, such as price pressure, shortened lead time, and competition (Kim & Davis, 2016;Lund-Thomsen, 2020).
Much attention has been devoted to how MNEs establish control in GVCs through levels of specialization, tailored relationships with the suppliers, and enhanced flexibility (Buckley, 2014). What is missing is a more systematic appraisal of the supplier's perspective. For example, Gooris and Peeters (2016) find that MNEs fragment global business processes across different product units to reduce the risk that suppliers misappropriate proprietary knowledge, but do not account for the suppliers' motives to participate in this business model. Relatedly, while Kano (2018) proposes a relational perspective on value distribution in GVCs that takes the business network as the central unit of analysis, the focus remains on how MNEs can enhance efficiency outcomes by adapting different social mechanisms.
Recent operations management and business ethics research has started questioning the prominent role of lead firms and the lack of agency of suppliers in promoting responsible conduct in value chain relations. Some of this work emphasizes how suppliers may proactively pursue strategies toward sustainability, which may not necessarily be a response to the strategies of the lead firms. De Marchi and Di Maria (2019) show how suppliers in the Arzignano leather cluster in Italy develop sustainability strategies to enhance competitiveness and achieve higher value creation. In developing countries, however, market power relationships and efficiency considerations take precedence (Jeppesen & Hansen, 2004). Lead firms remain the primary driver of supplier engagement in responsible conduct through ''hands-on'' standards-driven approaches, such as environmental upgrading (Golgeci et al., 2021). More visible MNEs are, thus, more likely to pressure their developing country suppliers to adopt environmental practices.
Unlike MNEs, suppliers, especially those in developing countries, tend to be less vulnerable to protests and activist actions, receive less diverse and intense stakeholder demands, and be less sensitive to demands for responsible business conduct (Fiss & Zajac, 2006). Therefore, these suppliers may not reap the ultimate economic benefits of responsible conduct, leading them to fear losing their cost-based competitive advantage (Goger, 2013) and assess the returns on investment in environmental upgrading as too uncertain. As Lund-Thomsen (2020: 1703) notes, ''CSR mainly appears to bring reputational benefits to brands/ retailers but few economic, social, or environmental benefits to suppliers''. Suppliers' advantages typically stem from their low-cost processes and environmentally lenient local regulators. Therefore, companies in these contexts often struggle to comply with the costly sustainability demands of their lead firms (Clarke & Boersma, 2017).

MNE monitoring in GVCs
When the economic interests of MNEs and their suppliers diverge, monitoring or auditing becomes relevant. Existing research assumes that MNE monitoring reduces the risk of irresponsible behavior by reducing information asymmetry (Gereffi & Lee, 2016). For example, MNEs can ensure effective governance through private standards that dictate what products will be made by whom and how (Lee, Gereffi, & Beauvais, 2012). Advances in communications, information technology, and transportation have been advocated to facilitate the monitoring of globally dispersed value chain activities (Buckley, Craig, & Mudambi, 2019).
Yet, MNE monitoring in GVCs is not straightforward. Gimenez and Tachizawa (2012) distinguish between ''assessment'' (including monitoring and auditing) and ''collaboration'' (working directly with suppliers to remedy problems). Assessment is difficult because suppliers' operations have limited visibility from the MNE's perspective (Narula, 2019), and monitoring is, therefore, inevitably limited by a deficiency of information. The traceability of potentially sensitive information in these relationships tends to be opaque (Lamming, Caldwell, & Harrison, 2004) due to the risk that information could be lost, omitted, or corrupted (Skilton & Robinson, 2009).
However, an even less well-understood question is what MNEs can and should do with the information they do retrieve about suppliers' conduct. There are limits to the improvements obtained by monitoring when failures are discovered and ''even after two failures, termination depended on caseby-case considerations and was only used in a minority of cases'' (Amengual & Distelhorst, 2019: 13). Thus, the connection between monitoring and responsible supplier behavior is complicated, challenging the appraisal of how MNEs can effectively rely on monitoring to reduce irresponsible GVC behavior. Accordingly, research suggests that standard internal monitoring mechanisms, such as codes of conduct, are insufficient to ensure responsible compliance across the GVC (Khattak & Pinto, 2018).

A MODEL OF CSR AND BARGAINING IN THE
GVC Our model addresses the research gaps presented above. We first present the ''baseline model'', where monitoring is prohibitively expensive (but subsequently relax this assumption and allow the MNE to monitor and react to any problems it discovers). We focus on a dyadic relationship between an MNE and a foreign supplier that provides an input (product or service) in return for a payment P. We envision the MNE and its supplier as a coalition that creates joint value by combining complementary resources, and we rely on the Nash Bargaining Solution (NBS) to characterize the value captured by each player as a function of its bargaining skills and outside option (see Nash, 1953).
Suppose that the MNE earns product market revenues of V, and the supplier incurs production costs of C, if they work together (see the Appendix for possible microfoundations of these parameters). Otherwise, each firm has an outside opportunity, worth O M for the MNE and O S for the supplier. We assume that V À C [ O M þ O S , so that there are some positive gains from working together. For example, if they terminate the relationship (or never commence it), the MNE must find a new supplier, and the supplier must find a new buyer. The level of competition on each side of the market will thereby determine their relative outside options; for example, if there are many, almost similar suppliers competing for the business of a few large MNEs, O M would be high and O S low. However, if the supplier is a large firm with unique capabilities (such as Foxconn), the MNE's bargaining position is weaker, and O M and O S can converge.
The two firms bargain over the gains from the relationship, with the MNE having bargaining ''bargaining skills'' or ''bargaining power'' of k. Hence, the MNE and the supplier get respectively a share k and 1 À k of the gains from the relationship. Before the gains are realized, however, the supplier chooses whether to be responsible or irresponsible. The latter choice could include, for example, polluting manufacturing practices, child labor, resorting to bribes, sweatshop practices, or wage-related violations. Aligned with research embracing the perspective of suppliers in GVCs (Goger, 2013;Lund-Thomsen, 2020), we assume that behaving irresponsibly provides private cost savings of D to the supplier (for instance, child labor or wage-related violations might be associated with lower production costs). Research has demonstrated that the opportunities to save costs by pursuing irresponsible actions are larger, suggesting a higher D, when the supplier operates in a less stringent regulatory environment (Clarke & Boersma, 2017).
The timeline can be depicted as in Figure 1, where the potential bargaining stages are also shown in dashed boxes.
Initially, the MNE and the supplier choose whether to work together. Each firm will participate only if its expected profit from agreeing to a deal exceeds the outside opportunity, and the deal will only be made if both agree. We assume that the MNE cannot discipline the subsequent behavior of the supplier via a comprehensive ex ante contract, including a clause that would punish the supplier severely when failing to honor this contractual obligation. This is a standard assumption in the economic literature on GVCs (Alfaro, Chor, Antras, & Conconi, 2019).
If a deal is made, the relationship begins and the supplier, at some later point in time, can choose (for now, unobserved by the MNE) to be responsible or irresponsible. Drawing on previous research (Feng, Lai & Lu, 2015), we treat this choice as a ''hidden action'' taking place after the MNE and the supplier have decided to work together. If the supplier is responsible, the two firms simply share V À C, based on their relative bargaining skills and outside options. If not, there is a probability x 2 0; 1 ½ that the irresponsible behavior will be revealed to the world through worker activism, the outside scrutiny of media, and NGOs (Berrone, Fosfuri, Gelabert & Gomez-Mejia, 2013). If the action is undetected, the firms bargain and share V À C þ D. This implies that the MNE also reaps part of the cost savings from irresponsible behavior because it increases the total economic value created by the relationship (see, for instance, Dukes, Gal-Or & Kannan, 2006). This can happen even if the MNE cannot directly observe or prove what the supplier is exactly doing, insofar as it will be able to form a belief about the supplier's irresponsible behavior and push down the negotiated price accordingly.
On the other hand, if the action is detected by the outside world, a more complicated bargaining situation arises, with two possible outcomes. First, the firms can choose to continue their relationship. However, the supplier will have to correct the irresponsible behavior and fix any associated consequences, thereby losing the cost savings D and, in addition incurring a cost of W in reflection of Gimenez and Tachizawa's (2012) collaboration process. For example, when IKEA's suppliers were found to be complicit in child labor, they were required to take corrective actions and ensure that the involved children could go to school with all costs paid by the supplier (Bartlett et al., 2006). In addition, both firms incur reputational damage from being revealed to be involved in an irresponsible supply chain; the MNE incurs B M and the supplier B S as the supplier's behavior is 'projected' onto the MNE as well (Foerstl, Reuter, Hartmann, & Blome, 2010). These reputational damages may depend on firm-, country-and industry-specific characteristics (e.g., Feng, Wang, & Kreuze, 2017).
Second, and alternatively, the two firms can terminate the relationship (Amengual & Distelhorst, 2019). In that case, each firm gets its original outside option, but now diminished by the reputational damage: The supplier earns O S À B S and the This means that the MNE, when disassociating itself from the irresponsible supplier, gets a ''discount'' on its reputational damage of F because it looks ''tough on irresponsible behavior'' and comes across as more righteous in the eyes of consumers (while the supplier gets no benefit from the termination of the relationship). Importantly, we assume that B F [ 0 , F\B M so that the MNE incurs some reputational damage even when terminating the relationship.
This suggests that some trade-offs are involved in deciding whether to terminate the relationship. Working together is better than termination if Hence, when irresponsible behavior is revealed, it would be rational for the firms to correct the problem if this inequality is fulfilled and terminate the relationship otherwise. Intuitively, it is a trade-off between the loss of the gains from the relationship on the one hand (V À C À O M À O S ), and, on the other hand, the costs of correcting the problem (W) and the incremental reputational costs of keeping the supplier (F).

Main Mechanism: Opportunistic Renegotiation
We build on the assumptions discussed above to solve the game and derive the payoffs in the various bargaining scenarios in which the MNE and the supplier can find themselves, as illustrated in Figure 2. The figure is drawn for the case in which the MNE keeps the supplier even if the irresponsible behavior is detected ( it suffers from greater reputational damage than the supplier does if the relationship is terminated after irresponsible behavior is detected (B F [ B S ) and the two firms have equal bargaining skills (k ¼ 0:5).
There are four cases in the figure. Starting from the left, in case (1) the supplier is responsible. The core of the bargaining is bounded by the willingness to pay of the MNE and the willingness of the supplier to accept a price, and the negotiated price P R splits the surplus among the two firms according to their bargaining skills (in this case, evenly).
In case (2), the supplier is irresponsible, but not detected. The MNE has the same willingness to pay as before, but now the supplier's costs are lower because it enjoys saving on its irresponsible actions. This means that the price falls to P N as the MNE is able to appropriate its share of the cost savings. Hence, both firms are better off than in scenario (1) and it is, thereby, also in the interest of the MNE to have an irresponsible supplier that is not detected. Essentially, cynical MNE managers are demanding a lower price from the supplier based on an unspoken assumption that the supplier is doing irresponsible things away from the spotlight of the world (and the MNE).
If the actions are detected by the outside world, the surplus is significantly reduced because of the reputational damage and the cost of correcting the problem. Most importantly, the revelation by the outside world changes the opportunity costs of both firms, leading to a potential renegotiation of the price after the revelation occurs. Notice that, if contracts were complete, the MNE and the supplier could ex ante negotiate an enforceable contract with a price that shares the total (reduced) surplus according to their bargaining skills, this price being P Ã D as shown in case (3), thereby eliminating the conflict of interest between the two firms.
The interesting case is, therefore, case (4), where contracts are incomplete and thus subject to renegotiation. This implies that when irresponsible behavior has been revealed, the supplier can threaten to walk away. Essentially, what the MNE is facing here is a hold-up problem, similar to Grossman and Hart (1986), but, in this case, caused by the detected irresponsible actions of the supplier rather than by investments. At the time of the renegotiation, some reputational costs can no longer be avoided -they are sunk costs and therefore subtracted from the outside opportunity. In other words, the two firms are no longer bargaining over the surplus but over the quasi-surplus, which is a bigger problem for the MNE than for the supplier (because B F [ B S ). Exploiting the MNE's vulnerability enables the supplier to renegotiate a higher price, P D instead of P Ã D . At that price, the MNE only gets a small slice of the total surplus, while the supplier gets by far the most, as can be seen in case 4 between the two dashed lines. This means that the MNE incurs a disproportionately high share of the costs of the irresponsible supplier's behavior, and the supplier is spared the worst consequences of its own actions. This creates a conflict of interest and can lead to the supplier being irresponsible, even though it would have been in the interest of the MNE to have a responsible supplier. 1 Focusing on case (4), our first proposition below captures the main theoretical mechanism behind our findings.
Proposition 1 After the supplier is detected by the outside world to have acted irresponsibly, it opportunistically renegotiates a higher price from the MNE, which suffers a hold-up problem, when: (a) The reputational benefit to the MNE from terminating the relationship is low, and (b) The MNE has limited bargaining power and experiences large reputational damage relative to the supplier.
The intuition behind Proposition 1 is the following. 2 First, if the MNE enjoyed high reputational benefits from terminating the relationship, it could use that threat to improve its bargaining position (condition a). This would reduce the opportunities for the supplier to benefit from irresponsible actions at the expense of the MNE. Second, in the hypothetical case in which the MNE has all the bargaining power (k ¼ 1), the gains from the relationship are always fully appropriated by the MNE, while the supplier always makes zero profits. This clearly would eliminate the hold-up problem. Thus, an important insight of Proposition 1 is that as soon as we depart from the conceptualization of GVCs as MNE-centric, we obtain novel and interesting insights pertaining to problems that would not occur if the MNE had very strong bargaining power. Third, condition (b) points to the crucial role of reputational damage due to irresponsible actions. This result speaks directly to the literature on CSR and MNEs. It shows that MNEs are more likely to be held up and suffer from opportunistic renegotiation by the suppliers when there is an asymmetry in reputational damage (low B S compared to B F ). This can be seen to reflect trends of MNEs being increasingly and aggressively targeted and blamed for the irresponsible practices of their suppliers (Bair & Palpacuer, 2012) and, at the same time, being more visible to stakeholders than these suppliers are. Importantly, if such reputational asymmetry is large enough, it can outweigh any inherent bargaining skill advantage that the MNE has.
The mechanism described above underpins the remaining propositions in the paper, as these pertain to strategies that the MNE can pursue to alleviate or avoid the hold-up problem. Henceforth, we will focus on the case in which the boundary conditions of Proposition 1 are fulfilled (the case in which the supplier has a stronger interest in responsible behavior than the MNE is discussed later). Finally, we also assume below that both the cost savings from irresponsible actions and the risk of detection are intermediate. We show in the Appendix that these conditions imply a conflict of interest, such that the MNE would like the supplier to be responsible, but the supplier prefers to be irresponsible.

Deterrence Failure and Social Branding as Commitment
When the relationship is valuable to the supplier, it will often be the case that the supplier will be responsible only if it expects the MNE to terminate the relationship when being detected. Unfortunately, when the relationship is valuable to the supplier, it will tend to also be valuable to the MNE (because they share the quasi-surplus according to their bargaining skills), and termination thus becomes a non-credible threat. Anticipating this, the supplier will be irresponsible, capitalizing on its ability to opportunistically renegotiate the price if something goes wrong. Empirical evidence suggests that this commitment problem may be a widespread issue. For example, Amengual and Distelhorst (2019) find that suppliers violating codes of conduct are no more likely to get terminated than other suppliers.
This also suggests, however, that if the MNE could somehow commit to terminating the relationship when the supplier is found to be irresponsible, it could better incentivize responsible behavior and thereby improve its own performance. For example, diversifying the MNE's supply base would improve the MNE's outside option, which in turn comes with the double benefit of strengthening the MNE's bargaining position to extract more value and serve as a credible commitment to terminate an irresponsible supplier, thereby influencing supplier incentives.
Another and more paradoxical way for the MNE to improve supplier incentives would be to increase its own vulnerability to reputational damage. Suppose that the MNE were to commit publicly to a zero-tolerance policy for irresponsible suppliers and publish detailed codes of conduct for suppliers, which would make it look hypocritical if it then did not follow up on those policies (Berrone, Fosfuri, & Gelabert, 2017). This would lead to an increase in the reputational damage that it suffers when working with an irresponsible supplier (B M ) by an amount Z. Assuming that these actions have no impact on B F , this would correspond to an increase in F (by implication of F ¼ B M À B F ). 3 Hence, even without considering the costs of the social branding itself (e.g., advertising expenses), this strategy would unambiguously worsen the payoff structure of the MNE, as some payoffs deteriorate and the rest are left unchanged. Yet, paradoxically, the indirect strategic effect could be positive. In particular, these actions could be a credible commitment device for the MNE to terminate the relationship, which, in turn, may change supplier incentives so that terminating the relationship never becomes relevant in equilibrium. This strategy is captured by Proposition 2: Proposition 2: Social branding investments that increase the MNE's reputational damage from working with an irresponsible supplier improve MNE performance by making the threat of terminating the relationship with an irresponsible supplier credible when: (a) The value of the relationship, including the cost of correcting the problem, is positive but small compared to the effect of the social branding investments, and (b) The supplier is powerful and resilient to reputational damage but has a low outside option.
The boundary conditions described in the proposition can be understood as follows. When the relationship is highly valuable and the cost of working to correct the problem is low, no reasonable amount of social branding would commit the MNE to terminate the relationship. On the other hand, if the value of the relationship is low and the cost of correcting the problem is high, the MNE will terminate the relationship no matter what; hence, no commitment device is needed. Therefore, it is only when the decision about terminating the relationship is a close call -because the relationship is moderately valuable, after considering the cost of correcting the problem -that the social branding strategies can tip the scale (condition (a)).
The supplier has more to lose when its bargaining skill is high and when its outside opportunity is poor (condition (b)). The intuition is that, if the relationship is terminated, the supplier then loses the opportunity to appropriate a good part of the value and, instead, gets its outside option. In this case, the threat of termination is stronger. Finally, the reputational damage of the supplier is a sunk cost when the irresponsible action has been revealed, and it, therefore, does not directly influence the supplier's cost of a terminated relationship. However, it does so indirectly by influencing the attractiveness of staying in the relationship: when the sunk reputational damage of the supplier is low, it has a better bargaining position and can extract more value, and, therefore, it has more to lose if the relationship is terminated (condition (b)).

MNE Monitoring
So far, we have treated the probability of the supplier's irresponsible behavior being detected as exogenous; that is, it depends on the scrutiny of the outside world, including media, activists and NGOs. However, monitoring in GVC relationships also results from the MNE's internal codes of conduct and specifically tailored activities (Locke et al., 2007). This suggests that detection probability is ultimately endogenous to the strategic choices of the MNE. In the following, we extend the baseline model with this possibility and explore how the MNE can alleviate (or, more surprisingly, exacerbate) the problem identified in Proposition 1.
To grasp the intuition behind our results, one needs to understand how the monitoring actions of the MNE overlap with the ''monitoring'' of the outside world. We capture the ''overlap between internal and external monitoring'' with a parameter 0 a 1. If a is high, it is very likely that the irresponsible behavior discovered by the MNE will be revealed at a later point in time by the outside world. In that case, we can say that the internal monitoring of the MNE is a substitute for the external monitoring of NGOs and media, and the primary advantage of monitoring for the MNE is to catch problems early so as to deal with them before they cause public scandal. On the other hand, if a is low, the irresponsible behavior discovered by the MNE will likely not be revealed after the fact. In that case, the internal monitoring of the MNE is additive to the external monitoring, and the primary advantage is that it increases the supplier's probability of being caught and, hence, may influence its incentives.
If the MNE does not discover the irresponsible action, the game proceeds as in the case without monitoring, where the world then detects the action with some probability. However, if the MNE discovers the irresponsible action first, it forms a belief about whether the irresponsible behavior will later be detected by the outside world and acts based on that belief. Details are omitted here but are available in the Appendix. There are now four possibilities: First, the MNE could, in principle, choose to ignore the problem. In that case, the world may or may not detect the problem later and we, therefore, revert to the baseline model, except that now an additional penalty d is added to the reputational damage of the MNE because it is revealed to have deliberately ignored irresponsible action in its GVC. It would therefore have been unambiguously better for the MNE not to monitor in the first place. Accordingly, we disregard this strategy in the following (as it never occurs in equilibrium).
Second, the MNE can work privately with the supplier to correct the problem at a cost of W, where 'privately' means that the MNE does not publicly disclose that it has discovered a problem. For example, Apple states that ''where we find issues, we work closely with the supplier on corrective action'' (ArsTechnica, 2020a), but at the same time has been known to avoid publicly announcing those issues until confronted by the press (ArsTechnicha, 2020b). If the MNE works privately with the supplier and the world does not discover the problem later on, the two firms share V À C À W, which is worse than if the supplier had chosen to be responsible from the beginning. However, if the world discovers the problem later on, the MNE will suffer brand damage of B M À R and the supplier of B S . Hence, the MNE does reap a reputational damage 'discount' of R for having been 'proactive' in working to correct the problem before it was discovered by the outside world, with the magnitude of R most likely related to the perceived credibility and sincerity of this work. However, at this stage, the MNE can no longer obtain F by terminating the relationship because it is revealed to have known about the problem and chosen not to terminate the relationship at an earlier stage. Similar to the base scenario, the MNE comes across as 'soft' on suppliers since it has not terminated the relationship, even after discovering irresponsible actions. We will show that this makes the MNE vulnerable to hold-up by the supplier.
Third, the MNE can announce the problem, sometimes referred to as supply chain transparency or disclosure (Doorey, 2011), while working with the supplier to correct it (Gimenez & Tachizawa, 2012). For example, Nestlé admitted to forced labor in its seafood supply chain in Thailand and commissioned a report to understand the problem and work with its suppliers (Guardian, 2015). When the MNE does this, it immediately suffers reputational damage of B M À R, unlike when the MNE works privately with the supplier and hopes to avoid reputational damage. On the positive side, announcing the problem immediately gives the MNE the power to threaten to terminate the relationship and reap the discount F (as described in the next scenario). Therefore, when announcing the problem, the MNE is in a better bargaining position with the supplier than when it tried to bury the problem and lost the opportunity to reap F.
Finally, the MNE can immediately and publicly terminate the relationship (Amengual & Distelhorst, 2019;Doorey, 2011). For example, Adidas terminated contracts with nine suppliers in 2013 due to non-compliance issues, including excessive working hours, non-payment of wages and benefits, and poor workplace safety conditions (CIPS, 2015). If it does so, this becomes immediately known to the outside world and the MNE incurs reputational damage of B M À F À R. Here, the MNE gets reputational damage discounts both for being tough on irresponsible suppliers and (still) for being proactive. We assume that B M [ F þ R, implying that the MNE suffers some positive brand damage even in this scenario, albeit the lowest level of damage it can achieve.

Optimal Response to Detection
Suppose that the MNE has detected a breach of social responsibility in its supplier, which has not yet been detected by the outside world (but may or may not be later on). We also assume that the relationship is sufficiently valuable so that the MNE will not terminate the supplier, leaving it with the choice between announcing the problem and working privately with the supplier to correct it. The choice between the two options depends on some conditions, as described in the following proposition: Proposition 3.1: After detecting irresponsible behavior by its supplier, the MNE chooses to announce it publicly when: • The overlap between internal and external monitoring is high, the reputational damages are small, the reputational discounts from being proactive and from terminating the relationship are large, and the bargaining strength of the supplier is high.
At first glance, it appears counterintuitive that the MNE would announce the problem at all. Common sense would suggest that managers would be hesitant to draw attention to problems that seemingly unnecessarily compromise the reputation of the MNE, if there is even a small chance that the outside world will not discover the breach later. However, the benefits of announcing go back, once again, to the bargaining situation between the MNE and the supplier and the potential for a holdup problem to arise.
Just like in the case of outside detection, the supplier can opportunistically renegotiate at some point in time after being detected by the MNE and beginning to work to correct the problem. Indeed, the MNE will look worse if it discovers the problem but waits to terminate the relationship, compared to announcing the problem and terminating the relationship immediately. The MNE is unable to make the supplier bear some of the incremental reputational damage from detection, which, therefore, becomes a quasi-surplus that is up for bargaining, leading to a hold-up problem. It can be shown that the incremental size of this quasisurplus is equal to F, the reputational discount from terminating the relationship. When the bargaining skill of the supplier is high, the hold-up problem is amplified, as we have seen earlier, and therefore it is more attractive to be proactive and publicly announce the problem. Finally, it is logical that it is better to announce if there is a high reputational discount from being proactive and the overlap is high such that the outside world is likely to discover the problem later.
However, even if it is not optimal to announce a problem at the time when it is discovered, it may be advantageous for the MNE to commit to such an announcement in advance. Suppose, for example, that the MNE could partner with NGOs (Huq, Chowdhury, & Klassen, 2016) and outside certification bodies in its monitoring efforts (De Marchi, Di Maria & Ponte, 2013), with the agreement that they would publicly announce any problems if detected -even if it would have been tempting for the MNE to not announce it if it had detected the problem itself. Such a commitment to public disclosure changes the incentives of the supplier, and the following proposition identifies the condition under which it enhances MNE performance: Proposition 3.2: Credible commitment to announcing the problem can increase MNE performance by influencing supplier incentives to behave responsibly when: • The cost savings from irresponsible action is moderate.
In the Appendix, we show that this scenario is more likely to happen if MNE monitoring is effective, the overlap between internal and external monitoring is low, and the likelihood of the supplier completely evading detection is low. Also, if the reputational damage to both parties from an announcement and the MNE's reputational discount of terminating the supplier are high, the supplier has more to lose from an announcement, and that makes the commitment more effective as a deterrent. Ultimately, Proposition 3.2 contributes to the literature on MNE-NGO partnerships, which has primarily focused on the unique capabilities of the NGOs (Huq et al., 2016), by suggesting that such partnerships may also have a credible commitment and incentivization benefit.
The Liability of Too Much Knowledge IB research has arguably regarded monitoring as the default solution to suppliers' irresponsible conduct (Buckley et al., 2019;Gereffi & Lee, 2016). In this section, however, we explore the complexity of the consequences of monitoring. We focus here on the scenario where the MNE will announce the problem if it discovers it (as defined in Proposition 3.1 above). In that case, the following proposition applies: Proposition 4.1: Monitoring an irresponsible supplier can influence the performance of the MNE negatively (even excluding the operational costs of the monitoring itself) when: • The overlap between internal and external monitoring is intermediate.
The surprising finding here is the possibility that monitoring can carry negative benefits, implying that the MNE would not profit from monitoring even if it were completely costless to do so. This finding contradicts a conventional agency perspective, in which the principal will always unambiguously benefit from having better information about the agent, and seemingly challenges the view that failure of information and knowledge exchange is the main barrier to successful supply relationships (Lamming et al., 2004). We now explain the intuition behind this paradox.
Suppose that the MNE suspects the supplier of being irresponsible and has to choose whether or not to invest in monitoring, with the aim of detecting the irresponsible actions before the outside world does. However, if the MNE is successful in detecting a problem, it finds itself in the following position. Ignoring it is too risky because d is high and it might suffer severe reputational damage if revealed later on. Terminating the relationship is too costly because it is valuable. Working privately with the supplier enables the supplier to renegotiate the price and hold up the MNE, because F is high. This leaves only the option of announcing the problem and working with the supplier. However, this is worse than never discovering the problem in the first place because R is low, so that the MNE gets relatively little credit for being proactive about the problem when it announces it, and W is high. Hence, it is expensive to correct the problem. In this case, the MNE managers will end up wishing that they had never found out about the problem, to begin with, questioning the wisdom of engaging in monitoring in the first place.
As discussed in the Appendix, two conditions must be fulfilled simultaneously for Proposition 4.1 to hold. 4 First, it must be rational for the MNE to announce the problem if it discovers it, requiring that the overlap between the monitoring of the MNE and that of the outside world is not too low (as described in Proposition 3.1). Second, the expected performance of the MNE when monitoring and announcing must be lower compared to not monitoring, requiring that the overlap is not too high -because in that case, the problem will be detected later on anyway and, then, the MNE has little to lose (and a reputational benefit to gain) by discovering the problem first. Together, this implies that the scenario here is more likely to happen if the overlap between the monitoring of the MNE and the outside world (a) is intermediate. Conversely, it can be shown that monitoring is most likely to be beneficial when a and R are high; the MNE is likely to detect a problem that would otherwise have caused a scandal, and it can then reap the reputational benefit of being proactive about it. It is also more beneficial when the MNE has an attractive outside option and does not suffer too much reputational damage. In that case, discovering the problem does not put it in a vulnerable bargaining position relative to the supplier.

Monitoring as a Deterrent or Encouragement
What we have described above can be seen as the direct effect of monitoring, where the only potential benefit comes from proactively managing the reputational damage of a presumed irresponsible supplier. However, it is also possible that there is an indirect, strategic effect in that the expectation of monitoring can influence supplier managers' actions. This effect is captured in the following proposition: Proposition 4.2: Monitoring the supplier can influence the performance of the MNE negatively, by incentivizing the supplier to be irresponsible when: • The MNE is expected to work privately with the supplier to fix the problem if discovered through monitoring.
As already shown in connection with Proposition 3.1, if the MNE is expected to announce the problem upon discovery, this can only have a positive effect on the incentives of the supplier to behave responsibly. Thereby, monitoring would discipline the supplier to act responsibly when a is high such that the MNE can be expected to announce the problem. However, if the MNE chooses to work privately with the supplier, the supplier may be better off because it can hold up the MNE and negotiate a better price, which may more than compensate for its share of the costs of correcting the problem. If monitoring improves the performance of an irresponsible supplier (and it does not influence the performance of a responsible supplier), this raises the possibility that monitoring will have the unintended consequence of incentivizing the supplier to be irresponsible.
This unexpected finding further proves the importance of moving away from the conceptualization of GVCs as MNE-centric and towards a dynamic bargaining perspective. It also challenges the idea of monitoring as an incentivizing device (Chen & Lee, 2017) or, at least, underlines the need for complementary strategies. In particular, it would be valuable for the monitoring MNE to achieve credible commitments to announce, as described in Proposition 3.2, or to terminate, as described in Proposition 2, as such commitments would avoid the problem identified in Proposition 4.2 and ensure that monitoring instead becomes an effective deterrent.

CONCLUSION AND DISCUSSION
By participating in GVCs, MNEs and their suppliers can exploit linkage economies that allow for higher levels of value creation (Gereffi & Lee, 2016;Mudambi, 2008). At the same time, GVC participation exposes MNEs to reputational risk as a result of potentially irresponsible social and environmental behavior at the supplier sites, such as child labor, unreported pollution, or labor rights violations (Barrientos et al., 2011;Stringer et al., 2016). In this paper, we have examined the conditions under which irresponsible behaviors emerge at supplier sites and how MNEs can reduce their occurrence. We study supplier incentives and motivation to comply with the MNE's CSR practices, how the bargaining power shifts between the MNE and its supplier as a consequence of differences in reputational damage due to irresponsible behavior, and the conditions under which it is in the MNE's best interest to accept, ignore or punish irresponsible behavior at the supplier site. We follow Hennart's (2009) encouragement to depart from an MNE-centric view of IB by specifically embracing how bargaining between MNEs and their suppliers shapes the emergence of irresponsible behaviors at the supplier sites due to asymmetries in power and reputation. Thus, we subscribe to his message that ''practitioners should take into account the goals and interests of owners of complementary local assets when setting up MNE strategy'' (Hennart, 2009(Hennart, : 1450.

Theoretical Contributions
We highlight three main theoretical contributions to existing IB literature. First, by analyzing a framework that allows for asymmetries in power and reputational vulnerability between MNEs and suppliers, we show that MNEs may be held up by suppliers when irresponsible actions in the MNE's GVC are revealed. Hence, our model is not one in which the MNE pushes a sustainability agenda because it is powerful, but, on the contrary, one in which it does so because it is vulnerable to reputation damage, and suppliers can exploit this vulnerability in opportunistic renegotiation. We show how these tensions arise and offer important impetus for future research to explore the consequences of power dynamics within GVC. We thereby demonstrate that power in GVC relationships can be conceptualized as ''diffused'' rather than ''direct'' as in the traditional MNE-centric view of these relationships (Dallas, Ponte, & Sturgeon, 2019). By showing that the MNE's fear of reputational damage can alter power dynamics in GVCs toward the supplier, we contribute to the less explored conceptualization of power as diffused in dyadic GVC relationships through the emergence of practices at the supplier sites that have potential repercussions on MNE reputation. Second, our model shows that monitoring is not a panacea and can even produce adverse consequences for the MNE. Specifically, monitoring an irresponsible supplier makes the MNE vulnerable to reputational damage if the outside world detects the supplier's wrongdoing, thereby improving the performance of the irresponsible supplier through opportunistic renegotiation, and, in turn, reducing the performance of the MNE. Most surprisingly, we show that monitoring can incentivize the supplier to behave irresponsibly. If the predictable outcome of detection is that the MNE will work privately with the supplier to correct the problem, the supplier will be in a position to renegotiate the contract. We demonstrate that monitoring is inherently an idiosyncratic measure and that the MNE should aim for its monitoring efforts to mimic outside world scrutiny when trying to manage reputational damage and to complement this scrutiny when using monitoring as a deterrent. We also show that monitoring is synergistic with commitments to announce problems and terminate irresponsible suppliers. This is an important contribution to IB research, which suggests that monitoring mechanisms such as advances in communications, information technology, and transportation can facilitate effective monitoring of globally dispersed value chain activities (Buckley et al., 2019).
Finally, we identify strategies MNEs can pursue to remedy the risk of irresponsible supplier behavior. By embracing the supplier perspective on CSR in GVCs more explicitly, we identify the conditions under which suppliers in GVCs are more likely to act irresponsibly. Consequently, this insight advances our understanding of the mechanisms underlying compliance/non-compliance of suppliers in developing countries (Lund-Thomsen, 2020). While prior research has largely assumed that irresponsible supplier behavior could be regarded as an exogenous element of the GVC (Stringer & Michailova, 2018), we demonstrate that CSR breaches are also endogenous to the behavior of the MNE. This is an important contribution, as it sheds light on how well-intended MNE actions can produce adverse consequences. On the other hand, it also suggests that MNEs can induce responsible behavior at the supplier sites with actions, like commitments to terminate an irresponsible supplier, or to publicly announce irresponsible action discovered through monitoring, that, by changing suppliers' behavior, can have indirect positive effects on MNE performance.

Limitations and Avenues for Future Research
Given the complexity of the relationship between MNEs and suppliers, we rely on some simplifying assumptions and make choices to restrict the scope of our analysis. We discuss below three areas that could be accommodated within our model and could be fruitful avenues for future investigation.
First, we have calibrated our model to analyze a conflict of interest where suppliers are the wrongdoers while MNEs are keen to implement sustainable practices in their GVCs. This is the scenario most often discussed in the literature (Locke et al., 2007;Lund-Thomsen & Lindgreen, 2014) and supported by the examples such as Apple, IKEA, Nike, and Zara. However, as shown in the Appendix, our model is sufficiently general to cover any other combination of preferences. First, both firms may prefer responsible action, in which case there is no problem to solve for the MNE, the supplier, the society, or IB scholars. Second, and more interestingly, we can have the 'opposite' conflict of interest, where the supplier is responsible while the MNE would prefer it to be irresponsible. This can happen if the supplier is vulnerable to reputational damage and has weak bargaining power. Because the MNE cannot commit to compensating the supplier for reputational risk but, in fact, will predictably exploit the supplier's vulnerability to extract more value -and because the supplier anticipates this -the MNE cannot prevent the supplier from being responsible. In that case, sustainability is achieved not because of the MNE but despite it. Finally, it is possible that both firms are interested in the supplier being irresponsible and that the supplier is willing to bear the reputational damage because it is well compensated. In that case, the only way that positive social outcomes can be achieved is by more direct government regulation (lower D), more scrutiny by NGOs and media (higher x), and/or more coherent stakeholder behavior (higher B S , B M , and lower F). This scenario captures the 'dark side' of global companies as it has been described in the literature on corporate wrongdoing (e.g., Giuliani, 2019). Recent EU legislation on corporate sustainability due diligence arguably reflects this view (Botwright & Verghese, 2022) as legislators aim to move such concerns out of the 'court of public opinion' into a legal compliance framework. We encourage researchers to explore the implications of different scenarios of irresponsible behavior and its policy implications. Second, our model can be extended to include specific investments by the MNE and the supplier to increase the value and decrease the cost, thereby endogenously expanding the gains from the relationship. 5 It can then be shown that the specific investment is higher when the supplier is expected to behave responsibly than when it is not. More interestingly, however, the investment might change the incentives of the supplier to behave responsibly and, thus, there could be ''overinvestment'' because a more valuable relationship increases the penalty of detection for the supplier. Relatedly, while we have employed a rather narrow focus on disclosure and monitoring as tools for MNEs to manage supplier relationships, other practices and arrangements, such as setting up new functional departments and working with third parties, are also relevant (e.g., Gong, Jia, Brown, & Koh, 2018). Future research could thus explore the implications of how MNEs can use a broader range of mechanisms to ensure sustainability in GVCs.
Finally, we have considered responsible and irresponsible behavior along a single dimension. MNEs and suppliers might behave responsibly in some activities and irresponsibly in others, or they may behave responsibly in one market and irresponsibly in another. While outside the scope of the current paper, we suspect that this consideration would not alter the fundamental trade-offs. However, findings will depend on how the different activities and the respective responsible and irresponsible behaviors interact with each other. In this context, a deeper examination of the fungibility of reputational damage would be important. For example, how does such damage spill over between independent firms that transact with each other, between subsidiaries or functions within the same firm, or across different geographies or industries? Can doing good in one context compensate for doing bad in another, or can the latter destroy the social brand built by the former?
Relatedly, while our model is dyadic between an MNE and a lead supplier, another plausible and arguably frequent scenario is represented by the case when multiple suppliers are subject to differing pressures from multiple MNEs that compete economically and reputationally. In our model, competition is implicitly captured in the outside options. For example, if there are relatively few MNEs competing for many interchangeable suppliers, O S will be low and O M will be high (and vice versa). However, future work could endogenize these options by explicitly modelling competition.
In conclusion, our model shows how MNEs are subject to strong information asymmetries, managerial dilemmas about how to respond when compliance breaches as detected, and difficult negotiations with the involved suppliers. A better understanding of these mechanisms can help the MNE reduce the risk of incompliant behavior, improve social goals, and manage reputational risks. While our model is theoretical, we provide several empirically testable propositions that capture our main mechanisms. Future research with data on dyadic interactions in GVCs could therefore explore, test, and extend these propositions. These data can be gathered through carefully reconstructing GVC relationships that companies increasingly disclose in their corporate reports and websites.
Journal of International Business Studies, and Organization Science. He is currently leading the research projects ''The Firm in the Informal Economy'' and ''Firm Formalization and Sustainable Development''.
Grazia D. Santangelo is Professor of Strategic and International Management at Copenhagen Business School and Visiting Professor at Henley Business School. Her research concerns the evolution of firms' internationalization process and how firms strategize on their intangibles -such as knowledge and reputation -across borders. She has published in leading academic journals, and her research has been recognized by international awards. She is Coeditor of Global Strategy Journal and an elected AIB and EIBA Fellow.

Open Access
This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Accepted by Jonathan Doh, Area Editor, 2 May 2023. This article has been with the authors for four revisions.