Reinforcing Regulatory Regimes: How States, Civil Society, and Codes of Conduct Promote Adherence to Global Labor Standards

,


2
to its "Code of Vendor Conduct" that requires compliance with domestic labor and environmental laws and forbids child labor, forced labor, and other abusive or discriminatory labor practices.Apple likewise contractually imposes a "Supplier Code of Conduct" on its suppliers and retained the Fair Labor Association, a nonprofit global monitoring organization, to enhance the monitoring of its supplier factories following activist pressure and media reports of widespread violations (Wingfield, 2012).
Codes of conduct and private social monitoring have gained significance beyond the corporate boardrooms of global brands.These governance tools have become the basis for a number of contemporary transnational regulatory regimes.In response to perceived limitations or failures of traditional government regulation (Büthe, 2010;Vogel, 2008;O'Rourke, 2003), transnational regulatory regimes that once focused on influencing state policies to raise global standards for labor, environmental, and human rights have instead begun to emphasize private nongovernmental means of achieving these ends.In 2000, the United Nations (UN) created the Global Compact, a set of widely accepted human rights, labor rights, and environmental sustainability norms to which corporations may voluntarily pledge their commitment (Lim and Tsutsui, 2012).For many years after, the UN explored how it might expand traditional notions of international law to make certain human rights commitments binding on TNCs as well as states.However, in 2008, the UN abandoned its attempts to legally bind corporations and instead adopted the "Protect, Respect, and Remedy" Framework, which clarifies that although states have binding duties under international law to "protect" human rights, corporations have the softer nonbinding "responsibility" to "respect" human rights.According to the UN Human Rights Council, this means that corporations should "act with due diligence to avoid infringing on the rights of others and to address adverse impacts that occur" (Ruggie, 2010:1).Corporate codes of conduct, and their enforcement through private social monitoring, are key governance mechanisms for implementing these and other transnational regulatory regimes (Shamir, 2005).
The shift to corporate governance as a primary mechanism for raising global standards for labor, environmental, and human rights springs from the perception that, in many contexts, TNCs may have more capacity than states do to influence the behavior of their supplier factories (Scott, 2010;O'Rourke, 2003).However, the shift from the state is far from complete.Several studies have demonstrated the ongoing relevance of states and domestic regulatory institutions to the implementation of private transnational regulatory regimes.Tim Bartley, for instance, details how local political and institutional conditions impeded the implementation of private transnational labor and environmental certification regimes in Indonesia (Bartley, 2010).Cesar Rodríguez-Garavito shows how state intervention shaped the resolution of disputes over code compliance in Mexico and Guatemala (Rodríguez-Garavito, 2005).Richard Locke and his colleagues have similarly documented distinct pathways of code implementation in Mexico, where private codes largely substituted for ineffective government regulation, and in the Czech Republic, where private codes built upon and complemented national labor legislation (Locke, Rissing, and Pal, 2012).
While studies like these have demonstrated the importance of institutional environments in shaping outcomes at particular firms, they have shed little light on the precise constellation of institutional conditions that fosters or predicts adherence to privately imposed transnational labor standards (Vogel, 2008;Büthe, 2010).The literature on private transnational regulatory regimes primarily addresses their adoption or diffusion (Wetterberg, 2011;Prakash and Potoski, 2006;Bartley, 2007aBartley, , 2007b;;Campbell, 2007;Hoffman, 2001).Those studies that have attempted to assess outcomes have been mostly qualitative, based on intensive case studies of a small number of firms in a single country, industry, or supply chain (e.g., Rodríguez-Garavito, 2005;Locke and Romis, 2007;Locke, Rissing, and Pal, 2012;Kocer and Fransen, 2009).
Our study seeks to fill these gaps in the literature that have been "crying out for more systematic analysis" (Büthe, 2010:9).First, it provides among the first broad-based, systematic comparative studies of compliance with private codes of conduct.Our study includes establishments in 11 industries in 43 countries.The scope of our data enables us to conduct the first evaluation of code adherence across a wide range of institutional environments, including public, private, domestic, and international regulatory regimes.Second, we assess not only the direct impact that various governance regimes have on code adherence, but also their interactive effects.Private transnational regulation is embedded in complex layers of governance emanating from a range of actors and sources of authority; it is crucial to understand how these institutions might enhance, neutralize, or undermine one another (Eberlein et al., 2012).Finally, we examine how code adherence is affected by institutions not only in the supplier-factory's country, but also in the buyer's country.In doing so, our work builds on an extensive literature that examines how local institutions affect factory labor conditions (Rodríguez-Garavito, 2005;Locke and Romis, 2007;Kocer and Fransen, 2009;Locke, Rissing, and Pal, 2012) and links this to a separate literature that describes how activist and consumer pressure has prompted TNCs to adopt codes of compliance and other corporate responsibility measures (Hendry, 2006;King, 2008;Chatterji and Listokin, 2007;Lenox and Eesley, 2009;Reid and Toffel, 2009;Jira and Toffel, 2013).Our study is among the first to examine whether the institutional conditions in developed markets affect labor conditions in TNCs' supply chains in developing markets.
Our findings substantially advance understandings of how institutional environments shape compliance with private transnational governance regimes.First, by demonstrating how states create domestic environments that promote adherence to the global standards embodied in codes of conduct, we advance the argument that states have an important role to play in transnational private regulatory regimes.We find that adherence to private labor standards is highest in countries that are closely tied to international institutions, as evidenced by their ratification of many labor treaties.We also find high adherence to codes in states that maintain high levels of press freedom.Second, we demonstrate that governance institutions are not fungible substitutes for one another, but rather they interact in complex ways, sometimes amplifying one another's effects on TNCs and their global suppliers.For instance, while the stringency of domestic employment laws in supplier-factory countries has no independent effect on private code compliance, it enhances the effect of treaty adoption.That is, in countries that have ratified more labor treaties, the existence of stringent domestic labor regulation promotes even greater code adherence.Similarly, while the presence in a country of many international nongovernmental organizations (INGOs) had no overall direct impact on factories' adherence to codes of conduct, it does amplify the protective effects of press freedom.Third, we demonstrate that there are institutional spillover effects from a buyer-country to a supplier-country.TNCs headquartered in countries where the issue of supply-chain labor abuse is particularly salient tend to work with suppliers that adhere more to the global labor standards embodied in codes of conduct.Taken together, these findings point the way toward building more effective private transnational regulatory regimes.

Corporate Codes of Conduct: Symbol or Substance?
Technological and economic developments in the late twentieth century, including trade liberalization, the turn to neoliberal government policies, and the globalization of production and finance, have rendered many states less willing or able to regulate corporate conduct in global markets.But these developments have not dampened the demand for more "rules of the game" to mitigate some of the harmful impacts of global capitalism (Jenkins, 2001:1).Instead, the locus of rule creation, diffusion, and enforcement has shifted from governments to private parties and to TNCs themselves, which are increasingly regulating their global supply chains through private means (George, 2007;Jenkins, 2001).
Codes of conduct are a common tool for private regulation of supply chains.All U.S.
Fortune 500 companies and thousands of other major TNCs have adopted codes governing labor and environmental standards and human rights in their supply chains (McBarnet, 2007).These corporate codes are imposed through supply contracts that stipulate not only transactional terms such as price, quantity, delivery, and quality, but also social, environmental, and human rights standards of production processes.The substantive features of these TNC codes are highly consistent, usually invoking broad international consensus standards such as those of the International Labour Organization (ILO).They also typically call for the supplier to comply with domestic labor, environmental, and human rights laws and specifically forbid practices such as child labor and prison labor, even if such practices are legal in the supplier's country or are prohibited only by unenforced laws (McBarnet and Kurkchiyan, 2007).Many TNCs have taken the additional step of hiring private social auditors to monitor their suppliers' adherence to their codes.
Contractually imposed codes of conduct hold both promise and peril as transnational governance mechanisms.On the one hand, some have argued that such codes can bring new voices into the regulatory process and can develop new mechanisms to motivate improvements for workers and the environment (Courville, 2003;Schrage, 2004).On the other hand, some fear that private codes of conduct will undermine the state's role in industrial policy-"leaving businesses to shape regulation according to their own liking" (Kocer and Fransen, 2009:238)and that they place a veneer of legitimacy over business practices that actually compromise rather than promote health, safety, and human rights (Banerjee, 2007).Such concerns have led many to dismiss corporate codes of conduct as "a farce" (Rodríguez-Garavito, 2005:212) or a "marketing device" that exacerbates global inequities of social power (Shamir, 2005:109).
Unfortunately, these polarized and largely theoretical assertions have shed little light on how the codes actually operate in practice and under what circumstances they are most likely to work.
Ascertaining the impact of corporate codes of conduct is necessary not only to understand private transnational labor regulation, but also to inform broader normative debates about corporate social responsibility (CSR).The ethics of engaging in CSR activities depend, at least in part, on whether such corporate interventions can mitigate the problems they seek to address (Margolis and Walsh, 2003).Our research aims to give companies the tools with which to assess where their attempts at private, contractual labor regulation are most likely to be successful.
Corporate codes of conduct adopted by TNCs bear all the hallmarks of hollow symbolism.First, TNCs often adopt codes of conduct under circumstances typically associated with a lack of commitment or capacity to implement them effectively, such as in response to pressure from nongovernmental organizations (NGOs), shareholders, or governments (Moon and Vogel, 2008;Schrage, 2004).But research has suggested that corporations are unlikely to follow through on "voluntary" commitments adopted primarily to deflect direct outside threats (Short and Toffel, 2010).The vast scale and rapid spread of code adoption suggests that much of it results from mimetic diffusion processes, as "managers seek to act in ways that are deemed appropriate by other managers and significant actors in their environment" (Campbell, 2007:958).Over the last two decades, prominent sources of corporate governance norms-such as business schools, leading business publications, and high-status corporations-have advanced the importance of socially responsible corporate behavior (Campbell, 2007), transforming CSR "from heresy to dogma" (Hoffman, 2001).In response, most major TNCs have adopted a code of conduct governing supply chain factories (Kolk and van Tulder, 2005).Although such mimetic pressures may increase the number of TNCs formally adopting codes of conduct (Wetterberg, 2011), research suggests that codes adopted under these conditions are highly likely to be symbolic gestures rather than substantive commitments (Meyer et al., 1997).
Second, codes of conduct are especially likely to be decoupled from corporate practices and priorities because the obligations they purport to impose conflict with TNCs' efficiencyoriented task objectives (Meyer and Rowan, 1977).The globalization of supply chains is motivated by price competition.Firms in more developed countries procure goods manufactured in less developed countries to take competitive advantage of low wages and regulatory costs.
Raising the cost of labor by adhering to standards imposed by codes of conduct seems to belie a fundamental logic of the global supply chain.
Third, the constituencies whose interests are protected by codes of conduct-primarily production workers in developing countries-are far removed from the powerful decisionmakers in most TNCs.Organizations are more likely to observe fragile, non-efficiency-related normative commitments when those commitments are supported by constituencies within the organization that have a reasonable amount of power as well as access and proximity to key decision makers (Rees, 1988;Selznick, 1992;Parker, 2002).Workers in far-flung global supply chain factories have no such power and have no such access to the corporate headquarters that impose these governance structures.
Finally, code violations are largely invisible to outsiders.Research suggests that corporations are more likely to live up to their voluntary commitments when their activitiesand transgressions-are visible to outside monitors (Short and Toffel, 2010;Toffel, 2008;Potoski and Prakash, 2005).The supply chain factories where code-of-conduct violations may occur are numerous and widely dispersed, their identities are often kept confidential by the brands that they supply, and they are often subject to little regulatory oversight in their home countries.It is therefore easy for both TNCs and their suppliers to hide code violations.In fact, several studies have demonstrated that corporate codes of conduct did little to improve factory conditions (Locke, Qin, and Brause, 2007;Esbenshade, 2004) and that there is often a significant gap between the rhetoric of voluntary codes and the reality of labor and environmental conditions (Esbenshade, 2004;Greenfield, 1997;Locke and Romis, 2007).
Despite such obstacles, there are reasons to believe that TNCs and their suppliers will, under some circumstances, adhere to their nonbinding commitments.First, literature suggests that even when auditing is adopted symbolically, it has governance effects that shape organizational practices and priorities.Orts (1995), for instance, describes how environmental auditing systems construct internal decision-making structures and routines that embed environmental impacts into managers' considerations.Research suggests that Sarbanes-Oxley's financial auditing requirements have changed corporate practices dramatically (Wagner and Dittmar, 2006;Schmidt and Raman, 2012) and that formal commitments to conduct internal audits can lead to improved regulatory compliance (Toffel and Short, 2011).In these types of regulatory design, auditing is not merely a technical tool for verifying objective facts, but an attempt "to re-order the collective and individual selves that make up organizational life" (Power, 1997:42) in a way that will "connect the inner workings of companies to wider public demands for control" (Power, 1997:65).Similar claims have been made for codes of conduct and the social auditing that often accompanies them.
Second, monitoring and other types of compliance pressure can come from many kinds of external source (Ayres and Braithwaite, 1992;McCaffrey and Hart, 1998;Parker, 2002).
Government pressure, in the form of either surveillance (Short and Toffel, 2010) or the prospect of more stringent public regulation (Lyon and Maxwell, 2007;Moon and Vogel, 2008), can prompt firms to live up to their voluntary commitments.INGOs can be a source of surveillance and political pressure prompting corporations to comply (Ayres and Braithwaite, 1992;Eesley and Lenox, 2006;Reid and Toffel, 2009).Research has suggested that unions (Pfeifer, Stefanski, and Grether, 1976) and the media (Dyck and Zingales, 2002) can also serve important monitoring functions.Many have posited that these civil society institutions play a key role in enhancing the efficacy of private transnational governance regimes (Fung, O'Rourke, and Sabel, 2001;Keck and Sikkink, 1998;Drahos and Braithwaite, 2001;Abbott and Snidal, 2009).
While these theoretical frameworks provide useful insight into the universe of factors that comprise successful transnational governance, they fail to explain the substantial variation in compliance with codes of conduct that has been observed.Rather than take the view that codes of conduct are either bogus or bona fide, we investigate the conditions associated with greater (or lesser) adherence to codes.
There is broad agreement that the efficacy of private monitoring depends, in part, on the characteristics of the institutional regime in which it is embedded (Vogel, 2010).As John Campbell (2007:963) notes, "[I]nstitutions are critical, especially if we are concerned with ensuring that corporations actually behave in socially responsible ways, rather than just pay rhetorical lip service to the issue."Private transnational regulation is embedded in multiple, cross-cutting institutional regimes.As discussed above, pressure to comply with legally unenforceable transnational norms can come from a variety of actors, from governments to private actors such as NGOs, unions, and consumers.To date, research on private transnational regulation has not attempted to address the multiple layers of governance at play.Below, we theorize how particular transnational and national government institutions, as well as civil society institutions in the supplier's and buyer's countries, encourage suppliers to adhere to the global labor standards embodied in privately imposed corporate codes of conduct.

Government Institutions and Private Code Compliance
Despite questions about the capacity of states to regulate transnational risks, research suggests that domestic regulatory environments help determine the effects of private regulatory efforts (Seidman, 2007).Locke, Qin, and Brause (2007), for instance, found better compliance with private labor standards among establishments in countries with strong rule-of-law norms.
By contrast, Kocer and Fransen (2009) found, in a study of Turkish factories, that private codes of conduct did little to promote workers' freedom-of-association rights because domestic law reinforced power imbalances between employers and workers.A country's labor regulation may derive from domestic law, international law, or some combination of the two.States administer domestic legal regimes governing employment relationships and workplace conditions.Some countries' labor and employment laws are more protective of workers than others.Domestic studies have demonstrated that companies are more likely to adopt self-regulation and implement it effectively when they are embedded in robust governmental regulatory regimes (Short and Toffel, 2008;Short and Toffel, 2010;Rivera, 2004;Rivera, DeLeon, and Koerber, 2006) and it is often argued that private transnational regulation is more effective when it is supported by credible state regulation (Vogel, 2010;Kocer and Fransen, 2009;Büthe, 2010).Factories embedded in more stringent domestic legal environments are likely to develop experience complying with protective labor laws and should thus find it easier and cheaper than other factories do to develop the organizational capabilities and normative orientation necessary to implement the transnational labor standards embodied in private codes of conduct.We therefore propose: Hypothesis 1 (H1): Suppliers in countries with highly protective labor law will exhibit greater adherence to labor codes of conduct than suppliers elsewhere.
In addition to enacting domestic law, many states have entered into binding international treaty regimes governing labor rights.The ILO is the intergovernmental organization and treaty regime governing global labor standards.It brings together stakeholders from governments, labor unions, and businesses to articulate broad principles and consensus standards for the protection of labor rights, including working hours, worker health and safety, child labor, and workers' freedom of association.These principles are contained in various ILO conventions that a governmental ILO member may elect to ratify as legally binding in its country.The ILO monitors its member governments' compliance with these conventions and provides compliance assistance to national regulators.
Treaties represent an important source of global norm diffusion and pressure to comply with international consensus standards (Lim and Tsutsui, 2012;Risse and Sikkink, 1999;Meyer et al., 1997).There is a vigorous and longstanding debate in the international relations and international law literatures about whether treaties are associated with the behavior of the governments that ratify them (Hafner-Burton and Tsutsui, 2005;Goldsmith and Posner, 2005;Hathaway, 2005;Kelley, 2007;Koskenniemi, 2011;Sachleben, 2006;Simmons, 2009).We address a different question-whether treaty adoption by host governments is associated with the behavior of private companies within those states.
An emerging literature has begun to examine this, but has produced no clear answers.
Hafner-Burton and Tsutsui (2005), for instance, suggest that human rights treaties may enhance the efficacy of human rights INGOs by supplying a set of norms that these organizations can draw upon to pressure governments to comply.They document separately the inefficacy of human rights treaties and the efficacy of human rights INGOs in reducing human rights violations, but their data do not demonstrate a relationship between treaties and INGOs.Lim and Tsutsui (2012) hypothesize that companies in countries with significant treaty obligations would be more likely to adopt the United Nations Global Compact, which asks companies to publicly declare their commitment to "universally accepted principles" regarding labor, human rights, the environment, and anticorruption, but the authors' results did not support that hypothesis.
In the context of global labor standards, we argue that labor treaty adoption by a host government reflects a domestic environment that encourages companies' adherence to private codes of conduct because the public and private governance regimes are mutually reinforcing.
Codes of conduct are private instruments that seek to provide a means of enforcing compliance with global labor standards like those embodied in ILO treaties (McBarnet and Kurkchiyan, 2007).TNCs often enforce these codes by engaging third-party auditors to monitor their suppliers' adherence, with implicit or explicit penalties for non-adherence.These private contractual arrangements provide a source of enforcement pressure lacking in many public international treaty regimes (Hathaway, 2005;Tsutsui and Wotipka, 2004;Hafner-Burton and Tsutsui, 2005).What private codes may lack, however, is legitimacy, being subject to criticism as undemocratic and illegitimate attempts to impose norms favored by powerful international actors (Bernstein and Cashore, 2007).Such concerns are less salient in countries that have publicly bound themselves to a similar set of norms through treaty ratification.A significant body of international law literature argues that treaties can be instruments of norm construction and diffusion (Keck and Sikkink, 1998;Risse and Sikkink, 1999;Chayes and Chayes, 1995;Goodman and Jinks, 2004).In short, we expect that international treaty obligations and private codes of conduct will be mutually reinforcing.We therefore propose: Hypothesis 2 (H2): Suppliers in countries that have ratified more labor treaties will exhibit greater adherence to labor codes of conduct than suppliers elsewhere.
Treaty obligations and domestic law do not operate in isolation from one another.In fact, because treaties typically have broad precatory language and weak enforcement mechanisms, their normative goals often can only be realized through the enactment of domestic law by ratifying governments (Koskenniemi, 2011).As Simmons (2009:12) notes, "[h]uman rights treaties matter most where they have domestic political and legal traction."Moreover, treaties can empower domestic political actors who support the global norms embodied in those treaties.
For example, "[t]reaties set visible goals for public policy and practice that alter political coalitions and the strength, clarity and legitimacy of their demands" (Simmons, 2009:12).In this way, treaties can shape the domestic political landscape in ways that promote the adoption of legislation to implement global standards.In countries that adopt labor treaties and operate strong domestic legal regimes governing labor, private labor codes reinforce the prevailing norms and legal enforcement mechanisms.Consequently, we propose: Hypothesis 3 (H3): Labor law and labor treaties are reinforcing mechanisms such that suppliers in countries that have both enacted highly protective labor law and ratified more labor treaties will exhibit greater adherence to labor codes of conduct than suppliers in countries with only one of these features.

Supplier-Country Civil Society Institutions and Private Code Compliance
The literature on self-regulation and corporate social responsibility envisions private third parties such as INGOs, unions, and the press as "a counterbalance to corporate power" (Campbell, 2007:958).A substantial literature suggests that INGOs can act as diffusers, monitors, and even enforcers of global norms, with the power to impose penalties on firms that fail to adhere to normative commitments (Vogel, 2010;Ayers and Braithwaite, 1992;Braithwaite, 1997;Sabel and Simon, 2012;Fung, O'Rourke, and Sabel, 2001).Lim and Tsutsui (2012) consider INGOs to be an important source of normative pressure compelling corporations to adopt the United Nations Global Compact and demonstrate that corporations are more likely to do so in countries with many INGOs.Seidman (2007) documents how the monitoring efforts of INGOs enabled the Rugmark certification system to police child labor in the Indian carpet industry.Many have observed the power of INGOs to "punish" companies that fail to adhere to global standards through "naming and shaming" campaigns (Fransen, 2012;Vogel, 2010;Soule, 2009;Seidman, 2007).Because INGOs can serve critical governance functions such as generating, monitoring, and enforcing norms, we propose: Hypothesis 4 (H4): Suppliers in countries with more INGOs will exhibit greater adherence to labor codes of conduct than suppliers elsewhere.
The media is another civil society institution that can provide regulatory pressure on factories.By monitoring and reporting on their behaviors, the press subjects companies "to the constant threat of public exposure" (Campbell, 2007:958).Companies' desire to maintain a positive media image (King, 2008) can exert strong regulatory effects on their behavior (Soule, 2009).Negative publicity can damage a company's reputation with consumers and diminish brand value (Vogel, 2010).Research has also shown that press coverage influences institutional investors and equity analysts more than many factors that receive significantly more attention in academic debates, including a company's board structure, auditors, and relationship with regulators (Low, Seetharaman, and Poon, 2002).The threat of substantial reputational penalties from negative media coverage of poor CSR practices can bolster companies' commitment to enforce CSR norms (Dyck and Zingales, 2002).Because this threat is greater in domains with more press freedom, we propose: Hypothesis 5 (H5): Suppliers in countries with more press freedom will exhibit greater adherence to labor codes of conduct than suppliers elsewhere.
A free press can also be leveraged by other civil society actors to increase regulatory pressures on companies.For instance, a free press can facilitate INGOs' ability to discover and publicize supply-chain labor abuse (Fransen, 2012), generating pressure on companies to improve their governance practices (Dyck and Zingales, 2002;Baron, 2003).Journalists and INGOs often work together to discover companies' practices (Dyck and Zingales, 2002), enhancing overall civil-society monitoring capacity.Research has shown that companies are more likely to concede to activists' demands on issues that receive a great deal of media coverage (King, 2008).For all these reasons, the press is central to the collective action strategies of many international labor NGOs (King, 2008;Braithwaite, 1997;Fransen, 2012).We therefore propose: Hypothesis 6 (H6): The positive influence of press freedom on suppliers' adherence to labor codes of conduct will be enhanced in countries with more INGOs.

Buyer-Country Civil Society Institutions and Private Code Compliance
TNCs are highly sensitive to brand reputation and consumer opinion (Vogel, 2010;Ruggie, 2003;Soule, 2009;Klein,2002).To protect themselves from negative publicity that might damage their image, TNCs often accede to the demands of mobilized consumers and activists to implement environmental and CSR measures such as codes of conduct (Fransen, 2012;King, 2008;Lenox and Eesley, 2009;Peretti and Micheletti, 2004).Often the mere threat of reputational damage, even without actual political mobilization, will cause a TNC to accede to such demands (Reid and Toffel, 2009;Soule, 2009).We argue that the potential for political mobilization in TNCs' domestic markets will similarly influence the way they implement the CSR measures they adopt.
Among the most important prerequisites to mobilizing anti-corporate activism is the framing of supply chain labor abuse in a way that generates a shared set of understandings and concerns about the issue among consumers and other stakeholders (Snow et al., 1986;Snow and Benford, 1988;Zald, 1996;King, 2008).The media, of course, play a key role in framing issues for a mass audience.Research has shown that exposure to an issue via the media can increase public concern about the issue (McCombs and Shaw, 1972;Page and Shapiro, 1989;Szasz, 1994) and that this shared concern can lead to political mobilization (McCarthy, Smith, and Zald, 1996;Klandermans and Goslinga, 1996).Accordingly, activists have used the media to amplify reports of supply-chain labor abuse, increasing the potential for mobilization and heightening consumer concern about the issue (Fransen 2012).Such concerns can also spill over to the company's employees, senior managers, and owners who do not want to be associated with an organization stigmatized by such reports.The potential for mobilization may be especially strong when the media coverage on supply chains depicts highly dramatic forms of abuse, such as child labor, which have "clear protagonists and visible effects" (McCarthy et al., 1996:295), making the issue highly salient.We thus expect that TNCs based in domains where the issue of supplychain labor abuse is particularly salient-and thus the risk of reputational damage and consumer mobilization is high-will be especially cautious in selecting and monitoring their suppliers' factories.We therefore propose: Hypothesis 7 (H7): Establishments supplying buyers in countries where the issue of supply-chain labor abuse is particularly salient will exhibit greater adherence to labor codes of conduct than suppliers elsewhere.

Data and Measures
We obtained, from one of the world's largest social auditors, a proprietary dataset including audit-level data for every code-of-conduct audit the firm conducted from 2004 through 2009.To avoid undue influence of atypical audits, we excluded audits from countries with fewer than 30 audits over the five-year sample.This resulted in a sample of 31,915 audits of 14,922 supplier establishments in 43 countries on behalf of 689 clients in 33 countries.The geographic composition of our sample is reported in Table 1.

Dependent Variable
To assess the extent to which factories comply with private codes of conduct, we calculated the number of labor violations from the social auditor database.This is the number of code-of-conduct violations the social auditor identified in the following areas: child labor, forced or compulsory labor, working hours, occupational health and safety, minimum wage, disciplinary practices, right of association, right to organize and bargain collectively, treatment of foreign workers and subcontractors, and dormitory conditions.Each of these domains includes a host of subcategories; for example, the occupational health and safety category includes items on emergency preparedness (7 items), fire safety (5 items), the canteen (2 items), toilets (8 items), and the work floor (8 items).The latter subcategory includes improper chemical storage, improper medical waste disposal, inadequate lighting, inadequate ventilation, lack of personal protective equipment, lack of chemical safety data sheets, unsafe electrical conditions, and unsafe machinery.To avoid undue influence of outliers, we winsorized (top-coded) this variable at the 99th percentile value of 28 violations.Figure 1 is a histogram of the number of labor violations.

Independent Variables
Factory-Country Employment Laws and Norms.For each country hosting a factory in the dataset, we calculated employment laws index, a measure of the country's labor protection and employment laws, obtained from Botero et al. (2004).This index incorporates legal protections of (1) part-time and fixed-term employment contracts, (2) maximum hours and premium wages for overtime, (3) severance pay and penalties, and (4) dismissal procedures.
To measure a factory country's proclaimed dedication to international labor protections, we used the number of labor treaties each country had ratified by each year.We obtained this data from the Database of International Labor Standards maintained by the International Labor Organization. 1 To reduce skew in our models, we use the log (after adding 1) of this count.

Factory-Country Civil Society Institutions.
We measure the extent of activist pressure in the country in which the audited factory operates as the annual number of international nongovernmental organizations in that country (Lim and Tsutsui, 2012).We obtained these data from the Yearbook of International Organizations (Union of International Associations, 2004Associations, through 2009)).Like Lim and Tsutsui (2012), we focus on international organizations categorized as Cluster I by the Union of International Associations, which includes federations of international organizations, universal membership organizations, and intercontinental membership organizations.To reduce skew, we use the log of the number of INGOs.
We measure the extent to which the audited establishment's country provides press freedom each year via the annual Press Freedom Index, produced by Reporters without Borders, 2 Because governmental and other institutions might influence the private sector differently in developed versus developing countries, we create a dummy variable, OECD, to distinguish Organisation for Economic Co-operation and Development (OECD) member countries from developing countries (Lim and Tsutsui, 2012). 4  To control for economic incentives that might affect the intensity of the audit, we obtained information from the social auditor on who had paid for each audit.We created three dummy variables (audit paid by buyer; audit paid by vendor, agent, or licensee; and audit paid by factory), coded "1" when each of these actors paid and coded "0" otherwise. 5o control for the possibility that an audit's emphasis might be affected by the use of third-party protocols, we created the dichotomous variable, third-party audit protocol.To control for the size and complexity of the factory being audited, we used the audit staff hours required for the audit,6 which we calculated as the product of the audit length (in hours) and the number of inspectors assigned to the audit, both of which were obtained from the social auditor's database.To reduce skew, we used the log (after adding 1) of audit staff hours in our models.

Methods and Results
We estimate the following model: where V ijkt refers to the number of labor violations identified in the audit of factory i in country j in year t on behalf of a buyer in country k.E jt refers to employment laws and norms in factory country j in year t and represents employment laws index and number of labor treaties.CSI jt refers to civil society institutions in factory country j in year t and represents international nongovernmental organizations and press freedom.B kt refers to buyer pressures in buyer country k in year t and represents issue salience in the buyer country.X ijt refers to all controls and includes GDP per capita, government effectiveness index, OECD, audit paid by buyer, audit paid by vendor, agent, or licensee (audit paid by factory is the omitted category), third-party audit protocol, audit staff hours, and full sets of industry dummies and year dummies.e ijkt is the error term.
We use centered versions of all variables that we subsequently use in interaction terms: employment laws index, number of labor treaties, international nongovernmental organizations, and press freedom.Because the dependent variable is a count variable that exhibits overdispersion, with a 5.9 ratio of the variance (41.0) to the mean (7.0), we estimate the model using negative binomial regression.Because many of our variables are measured at the countryyear level, we cluster standard errors by country.
Regression results of our primary model are reported in Column 1of Table 5 both as coefficients and as average marginal effects (AME).We begin by interpreting the controls.Audits of establishments in wealthier countries, as measured by GDP per capita, revealed fewer violations.Audits paid for by the buyer or by a third party (the vendor, agent, or licensee) resulted in more violations than audits paid for by the supplier (the omitted category).Audits of larger and more complex establishments, as indicated by requiring more audit staff hours, revealed more violations.
Our second model includes an interaction term between employment laws index and number of labor treaties (log).The significant negative coefficient on this interaction term supports H3 (Column 2: = -0.851, p < 0.01).To interpret this moderated relationship, Figure 2 plots average predicted values of number of labor violations associated with audits conducted in countries with various numbers of labor treaties.The declining slope of both lines indicates fewer code-of-conduct labor violations in countries with more labor treaties.The dashed line's steeper negative slope indicates that such violations decline more rapidly with labor treaties in countries that also have more robust employment laws (that is, above-median employment laws index) than in countries with less robust employment laws. 7The significant coefficient on the interaction term indicates a statistically significant difference between the slopes of these two declining lines.
We found no evidence that activist pressure, as measured by international nongovernmental organizations (log), was associated with fewer labor violations; our results do not support H4. Contrary to our prediction, the average effect is positive, an anomaly we explore below as an extension.
Returning to the results of model 1, significantly fewer violations were found at establishments in countries with more press freedom (Column 1: = -0.921, p < 0.01), which supports H5.A one-standard-deviation increase in press freedom is associated with 1.8 fewer violations (calculated as 1 SD * AME = 0.28 * -6.49).
Our third model includes an interaction term between press freedom and international nongovernmental organizations (log).The significant negative coefficient on this interaction term supports H6 (Column 3: = -0.678, p < 0.05).To interpret this moderated relationship, Figure 3 plots average predicted values of number of labor violations associated with audits conducted in countries with varying levels of press freedom.The declining slope of both lines indicates fewer code-of-conduct labor violations in countries with more press freedom.The dashed line's steeper negative slope indicates that such violations decline more rapidly with press freedom in countries that have more international nongovernmental organizations.The significant coefficient on the interaction indicates a statistically significant difference between the slopes of these two declining lines. 7For context, note the intercepts of the two lines in Figure 2.For countries with less employment law (belowmedian employment law index), the average predicted number of violations ranges from 11.0 with minimal labor treaties to just over half that amount, 6.5, when labor treaties are at the sample maximum.For countries with more labor laws (above-median), the corresponding range is 12.3 with minimal labor treaties to nearly one-third that amount, 4.2, with labor treaties at the sample maximum.
Significantly fewer violations were found in audits conducted for buyers in countries where child labor in supply chains was a more salient issue (Column 1: = -0.194, p < 0.01), supporting H7.A one-standard-deviation increase in the issue salience in the buyer's country is associated with 3.7 fewer violations (calculated as 1 SD * AME = 2.73 * -1.36).

Extension
Finding no evidence that activist pressure was directly associated with fewer violations, we explored whether this result might be cloaking heterogeneous effects.We reestimated our primary model but also included an interaction term between international nongovernmental organizations (log) and OECD.The results, reported in Table 5, Column 4, yielded a statistically significant negative coefficient on the interaction term (= -0.892, p < 0.01).To interpret this result, Figure 4 plots average predicted values of number of labor violations associated with audits conducted in countries at varying levels of international nongovernmental organizations (log).The declining dashed line indicates fewer labor violations in OECD member countries as the number of international nongovernmental organizations (log) increases, as H3 predicts.The slightly rising solid line indicates the opposite relationship in non-OECD countries, a relationship that warrants further research and might reflect a selection effect; that is, developing countries with more problematic working conditions might attract more INGOs.The significant coefficient on the interaction indicates a significant difference between the slopes of these two lines.

Discussion and Conclusion
Our findings substantially advance understandings of how institutional environments shape adherence to the global standards embodied in private transnational governance regimes.We find that the domestic institutional environments most hospitable to the imposition of global labor standards through private codes of conduct are those that have substantial connections to the international community, as evidenced by their many treaty obligations, and those that maintain high levels of press freedom.We find, further, that the effects of these institutions are amplified when they are layered with other public and private governance institutions.That is, treaties significantly shape domestic environments in their own right, but their effect on code compliance is enhanced in countries that also have a stringent employment law regime.
Similarly, while press freedom is highly predictive of code compliance, its effect is even more pronounced in countries with many INGOs.Finally, we find that institutions in the global buyer's country are associated with conditions in their suppliers' factories.TNCs in countries where the issue of supply-chain labor abuse is particularly salient work with factories that tend to be more compliant with the global labor standards embodied in private codes of conduct.Taken together, these findings make three key contributions to the literature and point the way toward improving working conditions in global supply chains and building more effective transnational regulatory regimes.
First, we go beyond the transnational regulation literature's consensus that states matter to show how they matter.Global labor standards were most successfully implemented in states that made significant formal and binding connections to the international community through their treaty ratifications and reinforced these connections by enacting stringent domestic employment legislation.By maintaining a free press, states also play a key role in promoting adherence to global standards.Thus, states are important not only for their exercise of traditional governmental functions, such as diplomacy and lawmaking, but also for their role in enabling civil society actors, such as the press, to exert regulatory effects.
Second, we demonstrate that the effects of different public, private, domestic, and international regulatory institutions are "interactive, not additive" (King, 2008:396) and that they are context-specific.The literature on regulation by civil society actors correctly observes that there are multiple actors capable of playing regulatory roles, especially in the international sphere (Abbott and Snidal, 2009;Keck and Sikkink, 1998;Ayers and Braithwaite, 1992;Braithwaite, 1997;Sabel and Simon, 2012).However, a subtext running through this literature is that these governance institutions can serve as substitutes or backstops for one another.Ayres and Braithwaite (1992), for instance, in their classic articulation of "tripartism," argue that when no government regulator is available to monitor and enforce compliance, public interest groups can act as a "proxy for the state" (Ayres and Braithwaite, 1992:132).INGO monitoring campaigns are often conceived as "an alternative to inadequate state enforcement" (Seidman 2007:124).Studies often portray INGOs stepping in to provide governance functions where the state lacks the will or capacity to do so (Chng, 2012).Our research challenges the implicit assumption that private governance regimes can serve as a substitute for state governance.In our findings, the presence of INGOs, for instance, had no independent effect on code compliance, but mattered only in states that maintained substantial press freedom.Moreover, the impact of INGO presence varied with the development levels of different countries.
Similarly, our findings highlight the frequently observed synergies between domestic and international legal regimes (Simmons, 2009;Kelley, 2007).The stringency of domestic employment law had no independent effect on code compliance, but it appears to be a crucial vehicle for implementing the global labor norms embodied in international treaty regimes.These findings support the arguments of those who have theorized that meaningful transnational regulation must weave together regulatory regimes operating "at many levels and across borders, deploying private rules, local practices, national laws, supranational forums, and international law in the interest of effective protection of workers and their rights" (Trubek, Mosher, andRothstein, 2000:1193).Our findings also underscore the importance of recent calls for more research on the interaction of governance regimes regulating transnational business activities (Eberlein et al., 2012).
Finally, we offer the first evidence suggesting that political pressures in TNCs' home markets might have a substantive impact on working conditions in foreign supply chains, an impact that goes beyond the symbolic rhetoric of code adoption.Although many studies document how stakeholder pressures in a buyer's domestic market can encourage it to adopt codes of conduct and other CSR activities (Fransen, 2012;King, 2008;Lenox and Eesley, 2009;Peretti and Micheletti, 2004), we are not aware of any studies that assess whether these measures were implemented effectively.Similarly, although studies have documented the effects that activist tactics such as shaming and boycotts have on outcomes that TNCs care about, such as stock price and earnings (Vasi and King, 2012;King and Soule, 2007), we are aware of none that investigate whether there is any connection between domestic institutional pressures on TNCs and working conditions in their supply chains.We show that domestic conditions that raise public awareness of supply-chain labor abuse in a TNC's home country are indeed associated with better compliance in that company's supply chains.This provides some support for private political approaches that seek to mobilize consumers and other stakeholders to directly target corporations.However, future research is necessary to determine whether particular anticorporate campaigns, rather than merely the conditions that might enable them, have produced better supply chain conditions.Taken together, these findings point the way toward building more effective private transnational regulatory regimes.We find that codes of conduct are most effective when they are embedded in states that have made binding domestic and international legal commitments to protect workers' rights and that have high levels of press freedom and nongovernmental organization activity.This research suggests the danger of assuming that one type of regulatory regime can substitute for another that has failed and reveals the critical importance of maintaining multiple, robust, overlapping, and reinforcing governance systems.

Figure 3 .Figure
Figure 3. Suppliers have fewer labor code violations in countries with more press freedom, especially in the presence of more international nongovernmental organizations

Table 2
reports the distribution of industries in our sample.Table3reports countries in each third of the sample distribution of employment laws index and number of labor treaties.Table 4 reports summary statistics and correlations.