Human Rights

American Political Science Review Vol. 103, No. 4 November 2009

doi:10.1017/S0003055409990116

Trade-based Diffusion of Labor Rights: A Panel Study, 1986–2002 BRIAN GREENHILL University of Washington–Seattle LAYNA MOSLEY University of North Carolina at Chapel Hill ASEEM PRAKASH University of Washington–Seattle

T his article investigates the nature of the linkages between trade and labor rights in developing countries. Specifically, we hypothesize that a “California effect” serves to transmit superior labor standards from importing to exporting countries, in a manner similar to the transmission of

environmental standards. We maintain that, all else being equal, the labor standards of a given country are influenced not by its overall level of trade openness, but by the labor standards of its trading partners. We evaluate our hypothesis using a panel of 90 developing countries over the period 1986–2002, and we separately examine the extent to which the labor laws and the actual labor practices of the countries are influenced by those of their export destinations. We find that strong legal protections of collective labor rights in a country’s export destinations are associated with more stringent labor laws in the exporting country. This California effect finding is, however, weaker in the context of labor rights practices, high- lighting the importance of distinguishing between formal legislation and actual implementation of labor rights.

C an international trade help improve the status of workers in developing nations? Specifically, are the labor rights of a given country influenced by the labor rights of its trading partners? In this

article, we examine whether such a “California effect,” in which key export markets exert upward pressures on outcomes in producer nations, does indeed exist. We examine how trade serves as a mechanism to dif- fuse norms and practices pertaining to collective labor rights from importing countries to exporting countries. We report evidence that the labor rights of a coun- try’s trading partners, rather than a country’s overall openness to trade, is a key determinant of labor rights outcomes. What matters for labor rights is not how much a country trades, but with whom.

The California effect, an idea formulated in the con- text of environmental issues (Vogel 1995), refers to a process by which economic exchange facilitates an expansion in the scope and stringency of regulatory standards in exporting economies. Rather than engen- dering regulatory races to the bottom, production for

Brian Greenhill is Ph.D. candidate, Department of Political Science, University of Washington–Seattle, Box 353539, Seattle, WA 98195 ([email protected]).

Layna Mosley is Associate Professor, Department of Political Sci- ence, University of North Carolina at Chapel Hill, 361 Hamilton Hall, CB 3265, Chapel Hill, NC 27599-3265 ([email protected]).

Aseem Prakash is Professor, Department of Political Science, Uni- versity of Washington–Seattle, Gowen 39, Box 353530, Seattle, WA 98195-3530 ([email protected]).

The authors thank Xun Cao, Judith Goldstein, Andrew Grant, Stephan Haggard, Anna Leander, Nita Rudra, and Michael Ward, as well as the editors and anonymous reviewers of the APSR, for their helpful comments and suggestions. Earlier versions of this ar- ticle were presented at the 2008 annual meetings of the American Political Science Association, the International Studies Association, and the International Political Economy Society. Aseem Prakash acknowledges financial support from the Center for International Business Education and Research, University of Washington. Repli- cation files for the results presented in this article can be accessed at http://dvn.iq.harvard.edu/dvn/dv/briangreenhill.

foreign markets with superior standards generates an upward trajectory in standards. This effect has been particularly apparent with respect to the diffusion of vehicle emissions standards across U.S. states (Vo- gel 1995).1 Indeed, Vogel coined the term “California effect” to describe the way in which states with a strong environmental agenda (e.g., California, or Germany in the European context) have been able to facilitate the diffusion of these environmental standards to other jurisdictions.

We examine whether there is something akin to a California effect for labor rights. In doing so, we en- gage with the broader literature on the role of inter- national economic and sociological networks in the transnational spread of a range of policies, including social security privatization (Brooks 2008; Weyland 2007), financial liberalization (Chweiroth 2007; Elkins Guzman, and Simmons 2006; Simmons and Elkins 2004), privatization of state-owned enterprises (Brune, Garrett, and Kogut 2004; Meseguer 2004), demo- cratic governance (Gleditsch and Ward 2006; Simmons, Dobbin, and Garrett 2008), and human rights (Greenhill forthcoming). In this literature, diffusion results from a variety of causal mechanisms, includ- ing international economic competition, direct pres- sure from intergovernmental organizations (IGOs), and learning among policy makers. Our theoretical ap- proach contributes to the broader diffusion literature by focusing on the role of global production networks (or supply chains) as the mechanism for transmitting labor practices from importing to exporting countries.

An important dimension of recent economic integra- tion is the globalization of production networks. Most corporations tend to source a large percentage of their inputs, components, and, in some cases, even finished products, from overseas suppliers. Alongside, multina- tional corporations have come under pressure from a

1 The controversy regarding California’s efforts to strengthen its auto emissions laws to respond to global warming, and the automobile industry’s initial opposition, further validates Vogel’s argument.

669

Trade-based Diffusion of Labor Rights November 2009

variety of directions—such as shareholders, nongovern- mental organizations (NGOs), and consumers—to en- sure ethical conduct and practices within their supply chains (Becker and Sklar 1999; Prakash and Potoski 2007; Spar and LaMure 2003). For many firms, the threat of political action by activist groups in importing countries concerned about buying goods from coun- tries that suppress labor rights, the ensuing media scrutiny, and the possibility of consumer backlash cre- ate strong incentives to pay attention to labor issues abroad. These incentives often exist even when multi- national firms use subcontractors, rather than directly owned production facilities, to carry out their overseas operations. For instance, during the past decade, firms in the apparel and footwear industries have faced in- creasing pressure to disclose their factory locations and to oversee workers’ rights in such locations (Bartley 2005). Similar dynamics exist within the carpet indus- try (Siedman 2007). Moreover, shareholder activism by ethically focused investment funds, as well as large in- stitutional investors such as Calpers, encourages firms to adopt codes of corporate social responsibility whose obligations extend to their overseas subsidiaries and suppliers (O’Rourke 2003). By establishing explicit corporate policies unilaterally or via their membership in voluntary labor codes, importing firms seek to ensure that their overseas subsidiaries and subcontractors re- spect labor rights.

Furthermore, we expect multinationals’ attention to labor rights within their supply chains to spill over to local firms as well. There is ample evidence that multi- national firms often bring their “best practices” to de- veloping nations, and that because of the sizable exter- nalities multinationals create in host economies, these practices are often adopted throughout the economy (Garcia-Johnson 2000; Moran, Graham, and Blom- strom 2005). In the context of China, Guthrie (2006) documents how local suppliers of multinational cor- porations, as well as local firms that want to join these international production networks, have initi- ated improvements in working conditions and workers’ rights (e.g., instituting formal grievance procedures) at the factory level. These micro-level changes, Guthrie suggests, are supported by macro-level institutional changes such as the establishment of the Labor Ar- bitration Commission; reformers in China have used globalization as an argument to push through domestic reforms.

Given these dynamics, we predict that countries ex- porting to destinations with higher levels of labor rights will have incentives to ratchet up their own labor stan- dards. Consequently, instead of observing a race to the bottom in labor standards, we should find a “trading up” in labor standards (Vogel 1995). Of course, the direction of change in labor standards (improvement or deterioration) is dependent on the labor standards in the major export destinations. Presumably, the impli- cations of the California effect would be less sanguine if countries showing disregard for labor rights absorbed the bulk of world exports. Given, though, that a sig- nificant proportion of exports from developing coun- tries are currently absorbed by developed countries,

and that these countries tend to have stronger labor rights protections, the existing trading context appears to create structural incentives for developing countries to improve labor rights outcomes.

In addition to establishing the existence of a California effect in an area other than environmental practices, our article contributes to the broader debate about the effect of trade on labor rights. Globaliza- tion critics assert that trade engenders a race to the bottom: lower labor standards (and, by implication, lower labor costs) provide a competitive advantage to exporting countries. Hence, those countries that want to gain a competitive edge in global export markets will have a strong incentive to refrain from providing legal protections for workers, or to flout those protections in practice (Collinsworth, Goold, and Harvey 1994). These critics predict that high levels of trade openness and export dependence will be associated with inferior labor rights, all else being equal.

In contrast, globalization optimists suggest that in- creased levels of trade will lead to gains in labor rights. Because trade openness is associated with economic growth and development, and because economic de- velopment can then spur political reform, openness will ultimately lead to better protection of labor rights. Oth- ers would argue that because higher levels of trade con- tribute to greater opportunities for interaction among states, trade can transmit superior human rights norms generally, and better labor standards specifically. Still others posit that, because foreign direct investors pre- fer locations with high levels of human capital, all else being equal, and because human capital development is strongly related to the more general protection of workers’ rights, foreign direct investment and human rights form a virtuous circle. Developing countries with better human rights practices attract more global pro- duction activity, and countries with more global pro- duction activity are more likely to experience longer- term improvements in human rights (Blanton and Blanton 2007). The globalization optimists therefore predict that, through a variety of mechanisms, higher levels of trade and global production will be associated with superior labor rights.

One could view this debate as simply an empirical one: all else being equal, is a country’s level of trade openness associated with better labor rights, worse la- bor rights, or no real difference in labor rights out- comes? Such a view, however, obscures an important flaw in the extant literature—namely, the incorrect the- oretical specification of how trade might impact labor rights. By focusing on a country’s aggregate level of trade openness, the existing literature tends to treat all trading relationships (importer–exporter) as the same: only the volume of trade, and not its destination, is taken to matter. But trade relationships are far from homogenous, and given the diversity in labor standards that exist across export destinations, a given exporter may well face conflicting pressures and signals from different importing countries. Some trade relationships may place downward pressures on labor rights, whereas others may motivate improvements. It is not simply how integrated into the global trading system a nation

670

American Political Science Review Vol. 103, No. 4

is that matters, but with which countries it trades. Thus, a textured analysis of trade that focuses on bilateral trading relationships is required to correctly specify how trade might affect labor rights in exporting coun- tries. Moreover, the level of trade between pairs of countries might in itself be dependent on existing la- bor rights, which is an issue we return to later in this article.

This distinction has broader implications for scholar- ship in political economy and for policy making. Treat- ing “trade globalization” in a disaggregated manner— that is, by carefully accounting for the policies and norms of each country’s trading partners instead of only focusing on total volumes of trade—improves our ability to understand the complex relationships be- tween economic globalization and labor rights. In this article, we argue that political scientists should pay close heed to the role of trading networks in trans- mitting standards, norms, and practices across coun- tries. In doing so, we join the “second image reversed” (Gourevitch 1978) scholars in emphasizing the embed- ded nature of the state and the susceptibility of domes- tic politics to international influences. As the broader literature on diffusion suggests, although domestic pol- itics and institutions are likely to be very important determinants of various labor-related standards, they most likely do not operate in isolation from external influences.

Moreover, internal and external factors may exert ei- ther upward or downward pressure on labor standards. Although we report evidence of a positive relationship between a developing country’s labor standards and those of its export destinations, other factors may ex- ert downward pressures on labor standards. Thus, the overall trend in labor standards is determined by the net effect of a variety of domestic and international variables such as changes in the level of economic de- velopment, population, and the effect of embedded- ness in other types of international networks. We find a positive marginal effect for the bilateral trade-related diffusion variable while observing a slight decline over time in the overall average labor standards of the de- veloping countries. Presumably, the California effect mitigates the downward pressure that other variables may be having on labor standards that, as we argue later in the article, is an important point that critics of economic globalization need to consider.

In the next section, we present our theoretical ar- gument regarding how the California effect dynamic operates in the case of labor rights. Next, we present our model and introduce our empirical approach. Then, we discuss the main results of our empirical analyses. In the final section, we conclude with suggestions for future research.

THE CALIFORNIA EFFECT IN LABOR RIGHTS

The California effect refers to the capacity of import- ing jurisdictions to affect the laws and practices of ex- porting jurisdictions. The effect obtains only when a

given importer accounts for a sizable share of an ex- porter’s market. Because California represents a large share of the U.S. automobile market, manufacturers located outside California—either in other U.S. states or in foreign countries—have been forced to adapt their production processes in order to ensure that their pro- ducts will meet California’s often higher environmental standards. In most cases, given manufacturers’ desires for economies of scale in production, these adaptations have led to improvements in the standards not only of the goods that are destined for the Californian market, but also of all goods manufactured by a given plant, and in some cases, across manufacturing plants focused on a given product.

A similar effect can be seen with respect to the dif- fusion of environmental standards within Europe. By the 1980s, Germany’s automobile industry had adapted to the strict emissions standards required for the ex- port of their cars to the U.S. market (specifically, to the California market, which accounted for half of Germany’s total auto exports to the US).2 In coali- tion with Germany’s powerful environmental move- ment, the auto industry began to lobby vigorously for the adoption of similarly strict standards in Germany and elsewhere in the European Community. Having made the necessary changes to their production fa- cilities, German car manufacturers did not want to compete with imported automobiles that fell short of these (somewhat costly) standards. Moreover, given the fact that Germany also represented an important import market for cars produced elsewhere, it too was able to facilitate the adoption of higher environmental standards among car manufacturers located in other European countries (Vogel 1995, Ch. 3). Trade ties therefore enabled the high regulatory standards de- manded by one particularly powerful market (the state of California) to be transmitted throughout the rest of the US, and, eventually, to Germany and the rest of Europe.

Scope Conditions

Vogel is careful to point out that the trade-related dif- fusion of stricter regulatory standards is contingent on a number of specific domestic and international con- ditions. Three separate factors account for the diffu- sion of California’s stricter environmental standards throughout the US. First, the very large size of the California automobile market gives the state enormous purchasing power. It is therefore able to make demands of its suppliers that would be unimaginable for smaller, less powerful economies. Second, California’s success in spreading environmental norms has depended to some extent on its domestic politics. California has a particularly powerful environmental lobby that has been responsible for pushing the state government to adopt high air quality standards in the first place. An- other market with a different constellation of political interests and institutions might adopt lower standards,

2 See Vogel (1995) p. 95.

671

Trade-based Diffusion of Labor Rights November 2009

or it might not even address the issue of environmen- tal protection. In such cases, trade-related mechanisms would not result in a ramping up of standards, and could even result in a lowering of standards.

Third, the California effect owes its success in part to the fact that California’s vehicle emission limits represent a product standard, and not a process stan- dard. The former refers to the physical features of a good, whereas the latter relates to the manner in which a good is created. Product standards are eas- ier to monitor, either within the domestic economy or at the border. Moreover, product standards have been deemed legitimate grounds for trade protection by international institutions, such as the World Trade Organization (WTO). Product standards therefore can confer a competitive advantage on manufacturers by making it easier for states to discriminate against goods that do not comply with the regulations (Vogel, 1995, 263). However, process standards are more difficult to monitor and enforce, and they have been deemed illegitimate by global trade institutions.

The extent to which we should expect a cross- national California effect to operate within the realm of labor rights depends on the extent to which these three scope conditions are satisfied. With respect to the first condition, the structure of contemporary in- ternational trade is such that, on average, countries with higher labor standards absorb the bulk of global exports, especially exports from developing countries. On average (and despite some diversity in standards among developed nations), labor laws and practices in developed nations are superior to those in developing countries, creating a clear possibility for a trade-based upgrading of standards. Given their market power, developed countries should have the capacity to en- courage improvements in the labor standards of their trading partners, provided they have the political will to do so.

As required by the second condition, a labor-related California effect requires that workers’ rights activists in importing states are sufficiently powerful to motivate importing firms to take notice of the practices of their subsidiaries and subcontractors. These interest groups must have the incentive to spend valuable political cap- ital pressuring consumers or corporations to limit im- ports from states with poor labor practices, or at least to “name and shame” companies that overlook this issue in their supply chains (Baron 2003; Spar and LaMure 2003). Human and labor rights activists and NGOs have an obvious interest in promoting higher labor standards abroad, but so too do other domestic interest groups such as labor unions that seek to protect domestic man- ufacturing industries from lower-cost imports. We can therefore expect coalitions to form around a common interest in restricting imports from countries with poor labor standards and to include diverse actors such as human rights groups, labor groups, and even certain industry representatives. Although these groups might not have the political power to push through import restrictions on goods produced using allegedly poor labor practices, they may—through their influence on consumer market dynamics—be able to force importing

firms to pay attention to labor practices in their supply chains.3

The third issue that emerges from Vogel’s discussion is the distinction between product and process stan- dards. Although the California effect may have been successful in transmitting product standards (e.g., auto emissions technologies) from one jurisdiction to an- other, its ability to successfully transmit process stan- dards faces various legal and practical obstacles. The WTO and its predecessor, the General Agreement on Tariffs and Trade (GATT), do not in general per- mit discrimination against imports based on process standards, except under the particular circumstances outlined in Article XX. The provisions of Article XX permit discrimination against imports produced using prison labor, but they do not otherwise allow for re- strictions related to producers’ violations of interna- tionally recognized core labor rights. Indeed, the WTO has tended not to address labor rights issues, in part be- cause of the GATT’s legal provisions against doing so and in part because calls for labor rights considerations can serve as a veil for developed country protectionism. Moreover, such discrimination would be very difficult to implement in practice.

Yet, despite the obstacles associated with process- based regulatory upgrading, more recent empirical work suggests that a California effect sometimes op- erates in the transmission of process-based standards. Prakash and Potoski’s (2006) study of the ISO 14001 environmental management (process) standard finds that levels of ISO 14001 adoption among exporting countries is strongly associated with the levels of adop- tion found among their export destination countries, even when controlling for a number of domestic and international variables. Furthermore, even without di- rect consumer pressure, firms might be motivated to pay close attention to process standards and the man- agement practices of their supply chains, given their public commitment to following socially responsible policies (Garcia-Johnson 2000; Prakash 2000). This suggests that, even in the absence of formal laws re- stricting the procurement of goods from countries with lax process standards, the desire of companies based in the importing countries to show evidence of a “clean” supply chain to their stakeholders can be sufficient to bring about the adoption of similarly high standards among their export partners. The question we explore is whether such process standard-based dynamics hold in the realm of labor rights as well.

More recent anecdotal evidence highlights the plau- sibility of a trade-based upgrading of labor standards. A range of multinational corporations, industry asso- ciations, and labor rights activists have encouraged the development and implementation of labor codes of conduct. These codes, based in the private sector, may supplement or substitute for labor laws and labor

3 A domestic legacy of protecting labor rights, and of political insti- tutions that encourage the protection of such collective rights (as in corporatist states such as Germany and Sweden), can also play a role here (see Hall and Soskice 2001; Huber and Stephens 2001; Mosley 2008).

672

American Political Science Review Vol. 103, No. 4

inspections in host economies. In 2003, a World Bank study estimated that 1,000 such codes existed (Smith and Feldman 2003). These codes vary in scope, strin- gency, and emphasis; some focus on health and safety issues; others focus on environmental issues; and still others emphasize payment of minimum or living wages. Many of them, however, take the core labor standards promulgated by the International Labor Organization (ILO) in 1998 as a key starting point. The freedom of association and the right to bargain collectively, as well as the elimination of child and forced labor, are therefore central elements. The codes also vary in their provisions for monitoring and enforcement. In more recent years, though, the general trend has been to- ward monitoring via third-party auditors (which could be private firms such as Ernst and Young or NGOs that work on labor rights issues); increasingly, these auditors are certified or trained by multistakeholder initiatives (representing industry as well as activists), such as the Fair Labor Association (Locke, Qin, and Brause 2007).

An important feature of many of these codes is that they apply not only to a firm’s directly owned facilities abroad, but also to a firm’s foreign suppliers. This is particularly important in labor-intensive indus- tries such as footwear and apparel, in which nearly all production is done via independently owned subcon- tractors. For instance, Nike’s list of supplier factories presently includes more than 700 locations, in 47 na- tions, and employing approximately 700,000 workers.4 In contrast, Nike has less than 25,000 direct employees. Nike’s behavior vis-à-vis labor rights issues parallels that of many multinationals: in the early to mid-1990s, the company came under scrutiny following various allegations of worker abuses at its supplier factories. After first denying responsibility for these problems, Nike attempted to position itself as an industry leader in the labor rights area, becoming the first firm in its in- dustry to disclose its list of supplier factors (Locke 2003, Locke, Qin, and Brause 2007). In addition, since 1997, Nike’s suppliers have been subject to various Nike- administered auditing programs to assess compliance with its corporate code of conduct. Nike now requires that factories supplying inputs (e.g., blank T-shirts) to Nike subcontractors be located in one of the approxi- mately 50 countries on the company’s list of approved production locations. Other firms in the industry, in- cluding Adidas, have followed Nike’s example.

Not surprisingly, there has been a long-running de- bate about the effectiveness of Nike’s monitoring pro- grams, as well as about the impact of corporate codes of conduct in general (Bartley 2005; O’Rourke 2003). But there also is evidence that, although many problems in supplier factories and exporting nations remain, Nike’s efforts have sometimes produced increased respect for various individual and collective labor rights, particu- larly in countries where government respect for rule of law exists (Locke, Qin, and Brause 2007). Similar

4 See http://nikeresponsibility.com/#workers-factories/main (acces- sed August 12, 2008).

patterns—in which U.S.-and European-based multina- tionals use corporate codes of conduct to influence conditions in exporting nations—have occurred in a variety of other industries, including carpets (where NGO Rugmark has worked with suppliers and retail- ers to limit child labor),5 soccer balls, and collegiately licensed apparel.6

In addition to changes in exporting country behavior that are promoted by private codes of conduct, some governments of importing jurisdictions have begun to link labor rights with market access. Although the for- malization of such linkages, via bilateral and regional trade agreements, has been a fairly recent develop- ment, it builds on a longer-standing concern with pro- duction processes in exporting nations. In the United States, the linkage between market access and labor rights dates to the 1984 Generalized System of Pre- ferences (GSP) Renewal Act. This act, amending the Trade Act of 1974, included a labor rights clause. Devel- oping nations’ eligibility for GSP status (a nonrecipro- cal set of trade concessions offered by individual devel- oped nations to developing countries) was to be based on, among other criteria, “whether a country was tak- ing steps to afford internationally recognized workers’ rights.” These rights were specified to include the free- dom of association and the right to organize, as well as individual working conditions (Compa and Vogt 2001). From 1984 to 2000, the U.S. International Trade Com- mission conducted approximately 100 labor-related re- views of GSP status, involving 42 countries. During this time, thirteen countries had their GSP preferences suspended, while an additional seventeen were placed on a “temporary extension with continuing review” list. Although GSP-linked trade comprises only a small per- centage of U.S. trade, some maintain that such reviews and suspensions have ripple effects: they can signal to other importing nation governments, as well as to multinational corporations (MNCs) and activists, that a given country has difficulty with labor rights (Compa and Vogt 2001). And, although GSP-related trade is small relative to total U.S. trade, exports to the US often comprise an important segment of a given low- income country’s trade.

More recently, U.S. government attention to labor rights issues has manifested itself via the inclusion of a labor side agreement in the North American Free Trade Agreement (NAFTA). In addition, the US-Jordan Free Trade Agreement (2000), the Cen- tral American Free Trade Agreement (CAFTA-DR, 2005), and the proposed US-Colombia Free Trade

5 Siedman (2007), however, finds little evidence that Rugmark, a voluntary program aimed at eradicating child labor in the Indian carpet industry, has significant effects on labor practices. She finds several problems with third-party monitoring-based systems in which the audited firms hire and compensate their monitors, creating in- centives for firms to hire the least stringent auditor. 6 Student-based activism relating to the licensing of collegiate ap- parel began in the mid-1990s in the US and has spread to a wide set of institutions. In 2008, several U.S. universities have discontinued their licensing agreements with Russell Corporation, following its closure of the Jerzees de Honduras manufacturing plant, allegedly due to a recent unionization effort by its workers.

673

Trade-based Diffusion of Labor Rights November 2009

Agreement also contain a range of explicit labor rights provisions. Perhaps most explicitly, the US-Cambodia Trade Agreement on Textiles and Apparel (1999–2004) offered Cambodia additional textile export access to the U.S. market in return for a guarantee that Cambo- dia’s national labor laws would be enforced. Oversight was conducted by ILO monitors in conjunction with the Cambodian garment manufacturers’ association, various NGOs, and U.S. buyers of Cambodian apparel.

Cambodia, which depended on textiles and apparel for 80% of its export earnings, used this agreement to attempt to create a niche as a “sweatshop-free” lo- cation (Chiu 2007). Although firms’ compliance with Cambodia’s relatively stringent labor code was not ab- solute, studies suggest a high level of provision of col- lective rights as well as individual working conditions (Abrami 2003; Chiu 2007). The program enticed sev- eral U.S.-based buyers to increase their imports from Cambodia (Fair Labor Association 2005). Cambodian firms that wanted to sell to U.S. buyers realized that they had to respect workers’ rights, and Cambodian workers realized that their firms would be monitored by various external agencies. Again, a domestic prefer- ence for labor rights in the import destination (in this case, promoted by U.S. labor unions and labor rights activists, and supported by Congress and the execu- tive) led to tangible improvements in labor rights in an exporting country.

Labor Laws and Labor Practices

Although our theoretical propositions apply to work- ers’ rights generally, our empirical analyses focus on the collective rights of workers. These rights—the freedom of association and the right to bargain collectively— are among the four “core” rights advanced by the ILO in its 1998 Declaration of Fundamental Principles and Rights at Work.7 The main conventions related to free- dom of association and collective bargaining have been ratified by the vast majority of governments, and the ILO’s declaration obligates (albeit with few direct en- forcement mechanisms) all members, even nonratifiers, to respect the core labor rights. In addition, standards governing workers’ collective rights provide the capac- ity to achieve more favorable outcomes in terms of wages, benefits, and working conditions (Elliott and Freeman 2003). The distinction between collective and individual rights can also be viewed as a distinction between enabling and protective rights (Rodrı́guez- Garavito 2005): the former include freedom of associ- ation and collective bargaining, and they can facilitate the achievement of improvements in the latter, which include working hours, minimum (and sometimes “liv- ing”) wages, and health and safety conditions. Indeed, although collective labor rights may not perfectly pre- dict actual working conditions, there is an established association between greater respect for collective la- bor rights and improvements in wages and working

7 The others include the elimination of forced labor, the prohibition of discrimination, and the elimination of all or some forms of child labor.

conditions (Aidt and Tzannaos 2002; Blanchflower and Bryson 2003; Flanagan 2006; Huber and Stephens 2001). For instance, several studies report the exis- tence of a “union wage premium” in a range of coun- tries. Other analyses consistently find an association between the use of formal labor market institutions to set wages and the dispersion of earnings (Abouharb and Cingranelli 2007; Freeman 2007). And, given that approximately one fourth of the world’s nonagricul- tural workers are members of labor unions—and that still more workers are covered by collective labor agreements—the potential for workers to act collec- tively is central to current debates regarding the impact of economic globalization on domestic outcomes.

In practical terms, focusing our analyses on workers’ collective rights leads us to examine outcomes that set a minimum floor for firms’ and governments’ behav- ior. This allows us to avoid the contentious questions of how to define an internationally acceptable level of working hours or wages or of how to assess the potential efficiency consequences of cash (individual) standards. We also avoid many of the difficulties as- sociated with measuring wages, benefits, and working hours in a cross-national context, which is particularly problematic with respect to developing nations (i.e., Daude, Mazza, and Marison 2003; Flanagan 2006).

In considering collective labor rights outcomes in exporting nations, we address two elements of workers’ rights: their legal provision and their practical imple- mentation. Although more recent studies tend to treat the legal and the practical elements within a single dimension (Mosley and Uno 2007; Neumayer and de Soysa 2005), we highlight the value of distinguishing the two. Indeed, in empirical terms, many countries that have strong legal protections for collective labor rights exhibit repeated violations of such rights in practice. Problems of compliance with laws are rife, particularly in nations that lack a well-established rule of law or the domestic regulatory capacity to oversee enforcement. Indeed, this domestic lack of regulatory capacity has been one justification for the development of private sector–based codes of conduct and monitoring systems (Bartley 2005). In addition, in theoretical terms, it is important to distinguish between a government’s at- tention to labor legislation and its practical implemen- tation of such rules.

In terms of the California effect, the issue is whether pressures emanating from importing countries bring about changes in an exporting country’s on-the-ground respect for collective labor rights, or whether they merely encourage exporting countries to legislate tougher labor laws that they either cannot or will not enforce. In other words, does the California effect ex- tend to tangible improvements in labor rights in ex- porting countries, or does it only encourage symbolic politics, where governments pass laws enshrining core labor rights, but fail to enforce them?

On the one hand, one could argue that the difficulties involved in monitoring labor rights in foreign jurisdic- tions make it unlikely that tougher labor standards will be implemented in practice. The fact that labor prac- tices constitute a process standard, as opposed to a

674

American Political Science Review Vol. 103, No. 4

product standard, makes monitoring compliance much more difficult. In the case of the environmental stan- dards that form the basis of the California effect, reg- ulators in California can easily determine whether im- ported vehicles comply with the state’s strict emissions standards. But consumers and firms located in the im- porting countries face severe information asymmetries about labor practices in exporting countries. They will find it difficult to verify claims about good or bad labor practices in exporting countries. Arguably, subsidiaries and contractors located abroad, and under pressure to control labor costs, will recognize and exploit their information advantages. Although exporting countries might enact new laws to assuage pressure groups in importing countries, they will have few incentives to actually enforce them.

On the other hand, consumers and NGOs based in importing countries are likely to recognize their in- formation disadvantages. Because only the exporting countries have the power to actually make improve- ments in their labor standards, consumers and activists might demand that the firms provide evidence that they are indeed respecting labor rights. By shifting the burden of proof to exporters, labor activists could suc- ceed in forcing the exporting firms to compensate the consumers for their information disadvantages. Con- sequently, multinational subsidiaries and subcontrac- tors in exporting nations will face pressure from actors higher up in their supply chain to demonstrate their compliance with national labor laws—and, where such laws are lacking, to offer evidence of practices that go beyond domestic legal requirements. This process has operated in several industries such as apparel, forestry, carpets, and coffee, where lead firms have required their suppliers to be certified as meeting international standards. Furthermore, ISO 14001 is a case in point whereby exporters and other firms in the supply chains are called on to demonstrate superior environmental practices (Prakash and Potoski 2006). In the case of labor standards, the pressure from importing jurisdic- tions to improve labor practices falls on exporting firms and their governments, which are keen to encourage exports.8

Nevertheless, the actual practices of manufacturers in exporting countries might be less likely to change than will be the formal laws and regulations govern- ing these practices. Presumably, the governments of exporting states that come under pressure—from the ILO, importing governments, or MNCs—to improve their labor practices will seek the often easier and more visible route of enacting new labor laws rather than changing their labor practices in a way that might

8 One could argue that ethical exporters might have incentives to pressure their domestic governments and trade associations to im- prove labor standards. After all, poor labor practices of one ap- parel exporter can generate negative reputational externalities for all apparel exporters. Indeed, the proliferation of codes of conduct sponsored by trade associations is partly to minimize such reputa- tional externalities. Moreover, respect for labor rights could provide a means for some suppliers and exporting nations to distinguish themselves in global markets, as in the case of Cambodia (Abrami 2003).

impose sizable political and economic costs. From the point of view of IGOs, providing technical assistance to improve national labor legislation is much cheaper than providing assistance to ensure compliance in prac- tice. Furthermore, the legal elements of collective labor rights are, in some ways, closer to product than to pro- cess standards: activists and MNCs need only look at the content of the legislation, rather than its implemen- tation in supplier factories, to determine compliance. Given these dynamics, we expect the California effect to be stronger in the context of labor laws than labor practices. Last, because changing industrial practices takes time, we also expect any California effect in labor practices to be operative after a greater time lag than a California effect in labor laws.

DATA AND EMPIRICAL MODEL

We model the relationship between each country’s col- lective labor rights outcomes and those of its trading partners using country-year data for 90 developing countries over the period 1986–2002. Countries from Africa, Latin America, the Caribbean, Asia, and the Middle East are included in our sample; we exclude the transition economies of Central and Eastern Europe, as well as those from the former Soviet Union. Omitted country-years from the developing regions are those for which data on one or more independent variables are not available. We focus on developing countries because they tend to have inferior levels of collective labor rights protection relative to developed countries. We do not include developed or post-Communist na- tions because the factors affecting labor rights out- comes in those countries are quite different from those determining outcomes in low- and middle-income na- tions (Mosley and Uno 2007).

If the race to the bottom argument holds, we can expect developed countries to begin to mimic the la- bor standards of developing countries, especially as economic globalization intensifies. If, however, the California effect holds, developing nations’ labor rights outcomes should converge with those of their export partners, for better or worse. Given the structure of trade during our sample period, we would expect to see developing countries mimicking the labor standards of developed countries. Consistent with the latter, throughout the 1986–2002 period, developing countries had a significantly lower average overall labor rights score than did the developed countries, for both laws and practices.

As discussed previously, our assessment of the California effect hypothesis treats separately the de jure and de facto aspects of a country’s level of respect for collective labor rights. The first measure, Labor Laws, gives an indication of the extent to which laws have been put in place to safeguard collective labor rights, such as the rights to organize, bargain collec- tively, and strike. The second measure, Labor Prac- tices, provides an indication of the degree to which labor rights are violated in practice. Both variables are derived from an aggregate measure of collective

675

Trade-based Diffusion of Labor Rights November 2009

labor rights employed by Mosley and Uno (2007). Im- portantly, this measure captures the extent to which restrictions are placed on workers’ freedom of associ- ation and collective bargaining—so-called “collective” labor rights, defined as one of the key elements of in- ternationally accepted core labor standards.

Following Kucera (2002), Mosley and Uno (2007) generated these data by conducting a detailed content analysis of reports on labor standards produced by three separate sources: (1) the U.S. State Department’s annual Country Reports on Human Rights Practices, (2) the ILO’s Committee of Experts on the Applications of Conventions and Recommendations (CEACR) and the Committee on Freedom of Association (CFA) re- ports, and (3) the International Confederation of Free Trade Unions’ (now part of the International Trade Union Confederation) Annual Survey of Violations of Trade Union Rights.

Kucera’s (2002) template records thirty-seven types of violations of labor rights, in six categories: freedom of association and collective bargaining–related liber- ties, the right to establish and join worker and union organizations, other union activities, the right to bar- gain collectively, the right to strike, and rights in export processing zones. In each of these broad categories, specific violations include the absence of legal rights, limitations on legal rights, and the violations of legal rights by government agents or employers. The coding scheme assigns a weighting to each type of violation, with more serious violations (e.g., general prohibitions on unions) weighted more heavily than others (e.g., a requirement of previous authorization in order for a union to join a confederation of unions). A full list of these categories and weightings is available in the Ap- pendix. Each country-year in the data set is assigned a score of either zero (no violations) or one (one or more violations) for each of the 37 categories of labor rights violations. For their analyses, Mosley and Uno (2007) add together the scores in each of these weighted cat- egories, deriving an aggregate measure of collective labor rights violations for each country-year. Possible scores on the aggregate labor standards indicator, then, range from zero to 76.5. In practice, however, no coun- try exhibits violations in every category of labor rights, and maximum scores are in the mid-30s. It is impor- tant to note that the overall measure, as well as the law and practice measures, indicate the total number of categories with violations in a given year, where categories that include more severe violations have greater weights. The measures, however, do not cap- ture the number of violations within each category— for instance, the number of workers dismissed on the basis of union membership—in a given country- year.

The separate Labor Laws and Labor Practices vari- ables were created by disaggregating this 37-point mea- sure into its separate law and practice components (see Table A1 of the Appendix). Typical “law” components of the scale include measures such as whether certain industrial sectors are allowed to impose limits on the right of workers to join unions or to strike (items 16 and 34), or whether workers need government approval

in order to engage in collective bargaining in the first place (item 25). In contrast, representative “practice” components of the scale include whether acts of vio- lence are reported to have been carried out against union leaders (items 1 and 2), or whether some firms make employment conditional on nonmembership in a union (item 9). As with Mosley and Uno’s (2007) overall measure of labor rights, each category of viola- tions is weighted in order to account for its severity.9 After disaggregating the overall labor rights scale in this way, our Labor Laws variable ranges from 0 to 28.5, whereas the Labor Practices variable ranges from 0 to 27.5.10 We have reversed the scale of both the La- bor Laws and Labor Practices variables so that higher values represent greater levels of respect for collective labor rights.

Our key independent variables are weighted aver- ages of the Labor Laws and Labor Practices found among a developing country’s export partners. We la- bel these variables Bilateral Trade Context: Laws and Bilateral Trade Context: Practices, respectively. These variables are constructed by taking the average Labor Laws or Labor Practices scores of each country’s ex- port destinations, and weighting these by the volume of goods exported to each of these destination countries in that particular year. Countries that export goods primarily to destinations with, for example, high scores on the Labor Practices variable will have high values on the Bilateral Trade Context: Practices variable, whereas those that send most of their exports to countries with poor labor rights performance will have lower scores on that particular variable. These variables capture the notion that it is the specific nature of a country’s trade relationships, rather than its overall level of trade open- ness, that underlie the trade-related causal mechanism for the transmission of labor rights. Data on export volumes at the dyadic level were obtained from the International Monetary Fund’s (IMF’s) Direction of Trade Statistics database. The calculation of this mea- sure can be represented as follows:

Bilateral trade contexti

= j∑

1

Labour rightsj × Exportsij

Total exportsi

where Bilateral trade contexti refers to the relevant bilateral trade context variable (either Law or Prac- tices), Exportsij represents the volume of exports sent from country i to country j, Labor rightsj refers to the relevant labor rights score (i.e., Labor Laws or La- bor Practices) for the destination country j, and Total exportsi represents the total volume of goods exported from country i to all its destinations. If a California

9 Note, however, that the weightings have little effect. For instance, for the overall labor rights scores, the correlation between the weighted scores and an index of unweighted scores is .89 for the global sample of countries and .87 for developing nations. 10 The Labor Laws and Labor Practice variables are only weakly correlated with each other (ρ = .27).

676

American Political Science Review Vol. 103, No. 4

TABLE 1. Illustrative Data on Values of Labor Laws and Bilateral Trade Context: Laws Variables for Randomly Selected Sample of Seven Developing Countries in Year 2002

Bilateral Trade Top Three Export Destinations (with Percentage Country Labor Laws Context: Laws of Total Exports Shown in Parentheses)

Egypt 12.50 23.10 US (21%), Italy (16%), UK (10%) Honduras 17.50 22.70 US (50%), El Salvador (12%), Guatemala (7%) Mauritius 18.50 25.60 UK (30%), France (22%), US (20%) Trinidad and Tobago 22.50 24.00 US (50%), Jamaica (8%), Barbados (5%) Congo, Democratic Republic 10.25 27.40 Belgium (68%), US (14%), Finland (5%) Thailand 21.00 22.00 US (20%), Japan (15%), Singapore (8%) Turkey 12.75 24.20 Germany (18%), US (10%), UK (9%)

effect does indeed operate with respect to labor stan- dards, we should find a positive relationship between each exporting country’s Bilateral Trade Context and its labor rights score in subsequent years. A sample of data illustrating typical values of Labor Laws and Bi- lateral Trade Context: Laws for several of the countries included in our data set is shown in Table 1.

In addition to our measures of Bilateral Trade Con- text, our model also includes two potentially impor- tant indicators of overall economic integration, each of which may influence labor rights in a “race to the bottom” or “climb to the top” dynamic. Total Trade is measured as each country’s total imports plus ex- ports as a percentage of its gross domestic product (GDP). This provides a measure of overall dependence on trade, and it represents the most frequently used metric of openness in extant literature.11 Including this measure allows us to test directly whether it is overall trade or the bilateral trade context that is more im- portant in the determination of a country’s labor rights outcomes. The second measure, FDI Inflows, reports the amounts of new foreign direct investment (FDI) received each year, expressed as a percentage of the country’s GDP.12 It therefore assesses the effect of di- rectly owned foreign production on labor rights out- comes. This measure tests the specific effect of directly owned foreign production (versus all production for ex- port by foreign or locally owned firms) on labor rights outcomes.13

Although Mosley and Uno (2007) report a negative relationship between overall levels of trade and labor rights, suggesting that increased dependence on trade causes countries to lower their labor standards in order

11 In our empirical analyses, we also replaced Total Trade with a measure of each country’s total exports as a percentage of its GDP. However, this does not lead to any notable changes in the estimated effect of our key independent variables, Bilateral Trade Context: Law and Bilateral Trade Context: Practices. 12 Data for both of these variables were obtained from the World Bank’s World Development Indicators. 13 Including a measure of total inward FDI stock—representing a country’s accumulated direct investment, rather than its investment in a given year—either alongside or in place of the FDI Inflows variable did not result in a significant change in the estimated effect of our key independent variables.

for their exports to remain competitive in the global economy, Neumayer and de Soysa (2006) find the op- posite. The relationship between FDI and labor rights is less contested: higher levels of inward FDI appear to be associated with improvements in labor standards. Possible causal mechanisms include the transmission of better labor standards from the parent company, or the ability of investing companies to pressure the host government into improving general levels of respect for the rule of law (see Mosley and Uno, 2007, 925–6 and references therein).

We also account for the possibility that participa- tion in certain types of trade agreements influences labor standards in developing countries. As discussed previously, Preferential Trade Agreements (PTAs) that govern trading relationships between major economies such as the European Union or the United States and many developing countries often contain a clause em- phasizing the parties’ commitment to protecting hu- man and/or labor rights. Over time, the inclusion of labor-related provisions in PTAs has increased. A more recent study of the impact of PTAs on countries’ lev- els of respect for physical integrity rights found that PTAs that incorporate human rights conditions can, in fact, lead to improvements in physical integrity rights, but only when the relevant human rights clauses have the potential to be effectively enforced (Hafner-Burton 2005).

Following Hafner-Burton (2005), we construct two dummy variables that indicate whether, in a given year, each country belonged to a PTA that includes either “hard” or “soft” human rights conditions. The variable Hard PTA is coded as “1” for each country-year in which the country belongs to at least one PTA that appears to make the trading relationship contingent on the maintenance of a high level of respect for hu- man rights. It is designed to indicate whether the state is subject to a set of enforceable human rights condi- tions. The variable Soft PTA, which indicates whether a state belongs to PTAs with unenforceable conditions, is coded as “1” for each year in which the country belongs to at least one PTA that makes reference to human rights in the text of the treaty, but does not appear to make the trading relationship conditional on a minimal human rights standard being upheld. These

677

Trade-based Diffusion of Labor Rights November 2009

variables were coded based on a content analysis of the PTA treaties available online from the WTO.14

Our models also include a number of controls for domestic-level determinants of labor standards. The Democracy variable reports the Polity2 democracy score for each country-year. This variable is the com- bined autocracy/democracy measure from the Polity IV database, which takes on values between −10 and +10, representing the most autocratic and the most demo- cratic countries, respectively. Previous studies have found democracy to be positively associated with col- lective labor rights (Mosley and Uno 2007; Neumayer and de Soysa 2006), as well as with improved human rights practices more generally. This is consistent with the idea that more democratic countries can better pro- tect the bargaining rights and rights to free association that constitute collective labor rights. We also include a measure of national income to account for the ways in which differing levels of economic development might affect labor standards. This measure, GDP per capita, is the log of the GDP per capita measure available from the World Development Indicators database.

In addition, we introduce a dummy variable to con- trol for the presence of civil war. Human rights vi- olations tend to increase dramatically when govern- ments face serious security threats (Hafner-Burton and Tsutsui 2005; Poe, Tate, and Keith 1999). Data on the occurrence of civil wars were obtained from the UCDP/PRIO Armed Conflict Database.15 The original four-point scale measuring civil war intensity has been recoded to create a dummy variable, where a value of “1” indicates the presence of a civil war of an “inter- mediate” or higher level of intensity. Our models also include measure of the log of population size, which has previously shown a strongly and statistically significant negative relationship with human rights practices (Poe, Tate, and Keith 1999; Richards, Gelleny, and Sacko 2001). Presumably, countries with larger populations will have a higher probability of violations occurring in various categories of labor rights, all else being equal. Population data were obtained from the World De- velopment Indicators database. A table of summary statistics for all variables is provided in Table A2 of the Appendix.

RESULTS

We report the results of our analyses, using ordinary least squares (OLS) regression with a random effects model, in Tables 2 and 3.16 We calculate robust standard

14 The various PTA treaties were downloaded from www. wto.org/english/tratop_e/region_e/region_e.htm (accessed July- August 2007). 15 Data were obtained from www.prio.no/cwp/ArmedConflict/ (ac- cessed August 23, 2005). 16 We use random rather than fixed effects because doing so allows for better consideration of within- and between-country variation. Our key independent variables, the bilateral trade context mea- sures, display significantly higher variation across countries (in a given year) than within countries (across years). Moreover, our data set includes a large number of countries (90) and a relatively small number of years (18). Both features suggest that a random effects

errors, clustered by country. Because changes in the la- bor standards of exporting countries can be expected to occur over an extended period of time, we first estimate the models using a one-year time lag for all indepen- dent variables. To assess the possibility of varying lag effects, we also reestimate the models with two- and three-year lags between the independent and depen- dent variables. Our model includes a lagged dependent variable because we expect labor practices in previous years to influence labor practices in subsequent years. Furthermore, the inclusion of lagged dependent vari- able mitigates the problem of serial correlation.

Labor Laws

Table 2 reports the results for the Labor Laws depen- dent variable. Our key independent variable, Bilateral Trade Context: Laws, has a positive and highly statisti- cally significant relationship with collective labor rights in all three models. The positive relationship indicates that high labor standards found among a country’s ex- port destinations are associated with improvements in the labor laws of the exporting country in subsequent years. Moreover, the effect of this variable becomes almost 50% larger when the time lag between the in- dependent and dependent variables is increased from one to three years. This provides support for the view that a California effect operates with respect to labor laws: countries that export goods to destinations with greater legal protections of collective labor rights will, over time, come to adopt similar legal protections.17

The coefficient for the Bilateral Trade Context: Laws variable presented in the third column of Table 2 sug- gests that for every unit increase in the average labor standards of a country’s export destinations, its own Labor Laws score will increase by approximately .3 units after a three-year lag, all else being equal. This implies a high degree of sensitivity to the labor stan- dards of a country’s trading partners, especially when one considers the fact that the Labor Laws variable has a standard deviation of only 2.3 units. The substantive and statistical importance of the Bilateral Trade Con- text also becomes apparent in Figure 1. The left-hand panel in Figure 1 provides an estimate of the extent to which changes over the observed range of values of Bilateral Trade Context: Laws affects the predicted Labor Laws score of a hypothetical country that has

estimator will be more efficient. However, random effects models do assume that the country-specific intercepts are uncorrelated with the country-specific covariates, which may be unrealistic. Fixed effects estimators, however, are particularly problematic when the model includes one or more relatively time-invariant covariates (Beck and Katz 2001). In this particular case, the GDP per capita (logged) and population (logged) variables tend to be stable over time. A fixed effects specification of our model, however, does produce statistically significant estimates of Bilateral Trade Context: Laws and Bilateral Trade Context: Practices when both the logged GDP per capita and logged population variables are dropped from the models. 17 Although the positive association between Bilateral Trade Con- text and Labor Laws could be consistent with either a “race to the bottom” or a California effect, we have strong theoretical reasons to believe that a California effect is driving these results. These are discussed in the next subsection, “Labor Practices.”

678

American Political Science Review Vol. 103, No. 4

TABLE 2. Effects of Regressing Labor Laws on Bilateral Trade Context: Laws, with Independent Variables Lagged by 1, 2, or 3 Years

1-Year Lag 2-Year Lag 3-Year Lag

Bilateral trade context: law .200∗∗ .242∗∗∗ .294∗∗∗

(.064) (.054) (.074) Total trade −.006 −.006 −.003

(.004) (.005) (.005) FDI inflows .003 .031 .025

(.026) (.035) (.025) Hard PTA .862∗ .940∗ .818

(.397) (.393) (.442) Soft PTA −.291 −.161 −.145

(.195) (.222) (.195) GDP per capita −.477∗∗ −.439∗∗ −.471∗

(.164) (.162) (.184) Democracy .038∗ .021 .020

(.019) (.018) (.020) Population −.382∗∗∗ −.391∗∗∗ −.355∗∗

(.086) (.096) (.104) Civil war .100 −.001 −.125

(.218) (.228) (.232) Lagged dependent variable .641∗∗∗ .629∗∗∗ .617∗∗∗

(.030) (.032) (.036) Constant 13.192∗∗∗ 12.143∗∗∗ 10.577∗∗∗

(2.362) (2.499) (2.780) N 1,424 1,338 1,252

FDI, foreign direct investment; PTA, Preferential Trade Agreement; GDP, gross domestic product. Notes: Robust standard errors are shown in parentheses. Significance levels are indi- cated as follows: ∗ p < .05, ∗∗ p < .01, ∗∗∗ p < .001.

FIGURE 1. Substantive Effects of the Bilateral Trade Context Variables

Bilateral Trade Context: Laws

L a

b o

r L

a w

s

10.0 15.0 20.0 25.0 28.5

10.0

15.0

20.0

25.0

28.5

Bilateral Trade Context: Practices

L a

b o

r P

ra ct

ic e

s

15.0 20.0 25.0 27.5

15.0

20.0

25.0

27.5

Note: The left-hand panel shows the effect that changes in Bilateral Trade Context: Laws has on the expected values of Labor Laws after a three-year lag for a hypothetical country where the values of all other variables are held constant at their median levels. The shaded area represents the 95% confidence intervals around the estimates. The right-hand panel does the same for Bilateral Trade Context: Practices. Note that the x axis in both graphs represents the range of values of Bilateral Trade Context found in the data set.

679

Trade-based Diffusion of Labor Rights November 2009

TABLE 3. Effects of Regressing Labor Practices on Bilateral Trade Context: Practices, with Independent Variables Lagged by 1, 2, or 3 Years

1-Year Lag 2-Year Lag 3-Year Lag

Bilateral trade context: practices .075 .073 .133∗

(.057) (.054) (.063) Total trade .004 .003 .002

(.004) (.004) (.004) FDI inflows −.053∗ −.0002 .015

(.025) (.021) (.026) Hard PTA .082 .115 .290

(.478) (.719) (.807) Soft PTA −.738∗∗ −.390 −.450

(.275) (.250) (.266) GDP per capita −.658∗∗∗ −.544∗∗∗ −.517∗∗

(.157) (.147) (.166) Democracy −.009 −.008 −.006

(.020) (.018) (.021) Population −.483∗∗∗ −.413∗∗∗ −.447∗∗∗

(.112) (.107) (.188) Civil war .070 −.263 −.521

(.311) (.289) (.339) Lagged dependent variable .549∗∗∗ .597∗∗∗ .580∗∗∗

(.027) (.026) (.029) Constant 20.419∗∗∗ 17.282∗∗∗ 16.629∗∗∗

(2.594) (2.700) (3.368) N 1,424 1,338 1,252

FDI, foreign direct investment; PTA, Preferential Trade Agreement; GDP, gross domestic product. Notes: Robust standard errors are shown in parentheses. Significance levels are indicated as follows: ∗ p < .05, ∗∗ p < .01, ∗∗∗ p < .001.

median values for all covariates included in the 3-year lagged model.

In contrast to our results for Bilateral Trade Context, the estimated effects of a country’s overall dependence on trade (Total Trade) and direct investment (FDI In- flows) are not statistically significant at the 95% con- fidence level.18 This result lends further weight to the argument that aggregate measures of economic glob- alization such as total trade and total FDI ought to be disaggregated for the purposes of assessing the impact of economic integration on labor rights.

Membership in PTAs with “hard” human rights con- ditions appears to be associated with greater legal la- bor rights protections. There exists a positive and sta- tistically significant relationship between the dummy variable indicating whether the state belongs to one or more “hard” PTAs and Labor Laws in subsequent periods. (The p values for the 1-, 2-, and 3-year lagged variables are .030, .017, and .064, respectively). How- ever, we do not see any statistically significant relation- ship between membership in PTAs with “soft” human rights conditions and Labor Laws in subsequent peri-

18 Interestingly, when we include a measure of overall capital ac- count openness that assesses the extent of controls on short- and long-term capital flows (Quinn 1997; Quinn and Toyoda 2007), we find that this variable displays a negative and statistically significant relationship with labor rights. The inclusion of this variable does not, however, significantly affect our estimates of the bilateral trade context variables.

ods. These results are broadly consistent with Hafner- Burton’s (2005) study of the relationship between PTA membership and physical integrity rights.

GDP per capita (logged) has a statistically signif- icant negative relationship to collective labor rights. This result is consistent with that of Mosley and Uno (2007), suggesting that among the sample of developing countries, those that are richer—and presumably more industrialized—are also the ones that are more likely to generate reports of violations of collective labor rights (but see Neumayer and de Sosya 2006). The Demo- cracy variable appears to have a positive relationship to Labor Rights in the first model, although this effect is not statistically significant after 2- and 3-year lags. Meanwhile, the dummy variable for the presence of a civil war suggests no statistically significant relationship to collective labor rights.19

19 We also test the hypothesis that the strength of civil society affects a country’s receptivity to international norms about labor standards (see, e.g., Neumayer 2005). We estimated a separate regression that included an interaction term between bilateral trade context and the number of international NGOs present in each state (which is assumed to serve as a proxy for the strength of civil society groups). The estimated effects of these interaction terms were not significant, suggesting that the presence of strong civil society groups does not augment the positive effect of bilateral trade context on labor rights. Instead, it appears that bilateral trade context has a more direct effect on a country’s labor standards.

680

American Political Science Review Vol. 103, No. 4

Labor Practices

When these models are estimated using Labor Prac- tices as the dependent variable instead of Labor Laws, the effect of Bilateral Trade Context is much weaker. As we report in Table 3, the estimated effect is positive in all three models, but is only statistically significant at the .05 level after a 3-year lag. (The p values for the 1-, 2-, and 3-year lags are .188, .173, and .035, respectively). An illustration of the substantive significance of this finding is shown in the right-hand panel of Figure 1.

The effects of the other independent variables are broadly similar, with the two most notable exceptions being the PTA variables. Here, we find no evidence of a statistically significant relationship between mem- bership in “hard” PTAs and Labor Practices, whereas membership in “soft” PTAs shows a negative rela- tionship to Labor Practices that is statistically signif- icant in the first model. One possible interpretation of this seemingly counterintuitive result is that member- ship in PTAs with enforceable human rights conditions has no statistically discernible effect on countries’ la- bor rights behavior in practice, whereas membership in PTAs with unenforceable human rights conditions merely provides a convenient way for abusive regimes to give the appearance of caring about labor rights. In other words, the negative relationship between Soft PTA membership and labor rights practices could be the result of a selection effect, whereby states with poor human rights records are more likely to choose to sign a PTA with human rights conditions that are ultimately unenforceable (see Hafner-Burton and Tsutsui 2005; Hathaway 2002; Vreeland 2008). Strategic and mis- leading signaling via such cheap talk is an interesting area of enquiry for future research.

Taken together, the results for the Labor Laws and Labor Practices models suggest that a California ef- fect operates with respect to the transmission of legal labor rights (de jure rights). Although labor-related outcomes in export destinations create incentives for exporting states to adopt tougher labor laws, the effect of these laws on ground-level outcomes (Labor Prac- tices) are weaker and discernible only after a three- year lag. Thus, making or changing laws may impose fewer political costs than actually enforcing them. This is a consistent with a mechanism of norm diffusion in which supply chain pressure leads exporting govern- ments to pass legislation improving labor rights, but does not force them to immediately take steps to im- prove labor practices on the factory floor. However, the statistically significant relationship between Bilat- eral Trade Context: Practices and Labor Practices after a 3-year lag suggests that the pressure on exporters to demonstrate compliance with labor laws, although weaker in magnitude, can eventually bring about real changes in behavior. Symbolic politics gets translated into concrete outcomes in a weaker fashion and only after a lag.

Figure 2 shows the global trend in both Labor Laws and Labor Practices over the 1985–2002 period, and compares the trends observed among the developing countries included in our sample with those of the Organisation for Economic Co-operation and Deve-

lopment (OECD) member states. Although the Labor Laws and Labor Practices of the developing countries are, on average, lower than the corresponding values of the OECD countries, both groups of countries appear to be experiencing a slight decline in labor standards over the duration of the sample period. This general decline is likely to be due in part to the negative effect that increases in GDP per Capita and Population— which have both increased steadily among the devel- oping countries in the sample—have had on labor stan- dards, given their negative coefficient estimates. Given that we report a positive marginal effect for both Bilat- eral Trade Context variables, exports to countries with higher labor standards should mitigate the overall de- cline in labor standards.

In addition, Figure 3 provides a more detailed picture of the distribution of the Labor Laws/Practices and the Bilateral Trade Context variables for the developing countries at the start and end of our sample period. As the median levels of the two types of labor standards have declined over the period, so too have the me- dian levels of the corresponding Bilateral Trade Context variables. It also reveals that a significant minority of the countries in the sample do in fact export goods to markets that, on average, have lower labor standards than their own. Does this mean that the positive coeffi- cient of the Bilateral Trade Context variables could be indicative of both a California effect and a “race to the bottom”? Although the positive coefficient indicates a general convergence between the labor standards of an exporting country and those of its export destinations— and does not distinguish between upward and down- ward pressures—we have strong theoretical reasons to believe that the convergence is likely to be the result of upward pressure on labor standards. For the reasons discussed at the beginning of this article, exporting countries will presumably face incentives to pass more stringent labor laws when exporting to countries that themselves have stringent labor laws. But the converse is unlikely to be true. When exporting to countries with less stringent laws, the exporter is unlikely to rewrite its laws (Labor Laws) to make them less consistent with international norms. Written laws tend to be sticky, es- pecially when it comes to sensitive issues such as labor rights. What is more plausible is that, when faced with competitive pressures, exporters may simply choose not to enforce their existing laws (Labor Practices) in an effort to reduce the cost of production (hence, the weaker result for Labor Practices as opposed to Labor Laws).

Robustness Checks

An important alternative explanation for the positive relationship between Bilateral Trade Context and labor standards could be that countries’ levels of trade with one another are to some extent conditional on prior levels of respect for labor rights. Although govern- ments might not be able to restrict imports on this ba- sis (given WTO rules), importing firms and consumers might simply refuse to purchase goods from countries with particularly poor labor standards. In this case, we would expect to see a positive relationship between

681

Trade-based Diffusion of Labor Rights November 2009

FIGURE 2. Trends in Labor Rights

1985 1988 1991 1994 1997 2000

0 5

1 0

1 5

2 0

2 5

Year

L a

b o

r L

a w

s

1985 1988 1991 1994 1997 2000

0 5

1 0

1 5

2 0

2 5

Year

L a

b o

r L

a w

s

1985 1988 1991 1994 1997 2000

0 5

1 0

1 5

2 0

2 5

Year

L a b o r

P ra

ct ic

e s

1985 1988 1991 1994 1997 2000

0 5

1 0

1 5

2 0

2 5

Year

L a b o r

P ra

ct ic

e s

Note: These boxplots indicate the distribution of the two measures of labor rights over time among both the developing countries included our sample and the Organisation for Economic Co-operation and Development member states (not included in our sample). The Labor Laws variable is shown in the upper panel, and the Labor Practices variable in the lower panel. The shaded boxes represent the interquartile range, whereas the vertical extent of the dashed lines show the full range of observed values of the variable for that year. The median values are shown as a black horizontal line.

Bilateral Trade Context and labor standards because of a selection effect: countries with high labor standards would tend to import goods from other countries with high labor standards, and countries with poor labor standards would be able to export to countries with similarly low standards.

To test whether this effect is responsible for the ob- served relationship between Bilateral Trade Context and labor standards, we reestimate the measures of Bilateral Trade Context using a series of instruments for exports. We first estimate export volumes at the dyadic level for 175 countries over the period 1980–

1999, using a modified version of the standard gravity model for international trade. The model specification is essentially the same as that used by Rose (2004) in his influential article on the relationship between WTO membership and trade. The only modifications we make are (1) to use the log of exports (instead of total trade) as the dependent variable,20 and (2)

20 Export data were obtained from the IMF’s Direction of Trade Statistics database. We added one unit (i.e., $1 million) to all export data to impose a lower limit of zero on the logged exports variable.

682

American Political Science Review Vol. 103, No. 4

FIGURE 3. Labor Rights and Bilateral Trade Context in 1985 and 2002

16 18 20 22 24 26 28

5

10

15

20

25

Bilateral Trade Context: Laws

L a

b o

r L

a w

s

AFG

DZA

AGO

ATG

ARG

BHS

BHR

BGD

BRB BLZ

BEN

BOL

BRA

BRN

BFA

BDI

KHM

CMR

CPV

CAF

TCD

CHL

CHN

COL

COM

ZARCOG

CRI

CIV

CUB

DJI

DMA DOM

ECU

EGY SLV

GNQ

ETH

FJI

GAB

GMB GHA

GRD

GTM

GIN

GNB

GUY

HTI

HND

IND

IDN

IRN

IRQ

ISRJAM

JOR KEN

PRK

KOR

KWT

LAO

LBN

LBR

LBY

MDG

MWI

MYS

MDV

MLI

MRT MUS

MEX

MNG MARMOZ

MMR

NPL

NIC

NER

NGA

OMN

PAK

PAN

PNG

PRY

PER

PHL

QAT

RWA

STP

SAU

SEN

SYC

SLE

SGP

SLB

SOM LKA

KNA

LCA

VCT

SDN

SUR

SYR

TZA

THA

TGO

TTO TUN

TUR

UGAARE URY VUT

VEN

VNM

WSM

ZMB

ZWE

16 18 20 22 24 26 28

5

10

15

20

25

Bilateral Trade Context: Laws

L a

b o

r L

a w

s

AFG

DZA

AGO

ATG

ARG

BHS

BHR BGD

BRB

BLZ

BEN

BOL

BRA

BRNBFA

BDI

KHM

CMR

CPV

CAF

TCD

CHLCHN

COL

COM

ZAR

COG

CRI CIV

CUB

DJI

DMA DOM

ECU

EGY

SLV

GNQ

ETH

FJI

GAB GMB

GHA

GRD

GTM GIN

GNB

GUY

HTI

HND

HKG

IND

IDN

IRN

IRQ

ISR JAM

JOR KEN

KIR

PRK

KOR KWT

LAO

LBN LBR

LBY

MAC

MDGMWI

MYS

MDV MLI

MRT

MUS

MEX

MNG MAR

MOZ

MMR

NRU

NPL

NIC

NER

NGA

OMN

PAK

PAN

PNG

PRY

PER

PHL

QAT

RWA

STP

SAU

SEN

SYC

SLE

SGP

SLB

ZAF

LKA

KNA

LCA

VCT

SDN

SUR

SYR

TZA

THA

TGO

TON

TTO

TUN

TUR

TUV

UGA

URY

VUT

VEN

VNM

WSM

YEM

ZMB

ZWE

16 18 20 22 24 26 28

5

10

15

20

25

Bilateral Trade Context: Practices

L a b o r

P ra

ct ic

e s

AFG DZAAGO

ATG

ARG

BHS

BHR

BGD

BRB BLZ

BEN BOL

BRA

BRN

BFA

BDI KHM

CMR

CPV

CAF

TCD

CHL

CHN

COL

COMZAR

COG

CRI

CIV CUB

DJI DMA

DOMECU

EGY

SLV

GNQETH

FJI

GAB

GMB

GHA

GRD

GTM

GINGNB

GUY

HTI

HND

IND

IDN

IRN

IRQ

ISR

JAM

JOR KEN

PRK

KOR

KWT

LAO

LBN

LBR

LBY

MDG

MWI

MYS

MDV MLI

MRT

MUS

MEX

MNG

MAR

MOZ

MMR

NPL

NIC

NER

NGA

OMN PAK

PAN

PNG

PRY

PER

PHL

QATRWA

STPSAU

SEN

SYC

SLE

SGP

SLB

SOMLKAKNA LCAVCT

SDN SUR

SYR

TZA

THA

TGO

TTO

TUN

TUR

UGA

ARE

URY

VUT VEN

VNM

WSM

ZMB

ZWE

16 18 20 22 24 26 28

5

10

15

20

25

Bilateral Trade Context: Practices

L a b o r

P ra

ct ic

e s

AFG

DZA

AGO

ATG

ARG

BHS

BHR

BGD

BRB

BLZ

BEN

BOL

BRA

BRN BFA

BDI

KHM

CMR

CPV CAF

TCD CHL

CHN

COL

COM

ZAR

COG

CRI

CIV

CUB

DJI

DMA

DOMECU

EGY

SLV

GNQ

ETH

FJI

GAB

GMB

GHA

GRD

GTM

GIN

GNB

GUY

HTI

HND

HKG

IND

IDN

IRN

IRQ

ISR JAM

JOR

KEN

KIRPRK

KOR

KWT

LAO LBN

LBR

LBY

MAC

MDG

MWI

MYS

MDV

MLI

MRT

MUS

MEX

MNG

MAR

MOZ

MMR

NRU

NPL

NIC

NER

NGA

OMN

PAK

PAN

PNG

PRY

PER

PHL

QAT

RWA

STP

SAU

SEN

SYC SLE

SGP

SLB

SOM

ZAF

LKA

KNA

LCA

VCT

SDN

SUR

SYR TZA

THA

TGO TON TTO

TUN

TUR

TUV

UGA

ARE

URY

VUT

VEN

VNM

WSM

YEM

ZMB

ZWE

Note: These scatterplots show the association between the two measures of bilateral trade context and labor standards at the start and end of the sample period. Each country is identified by its three-letter ISO 3166 code. The vertical positions of the points have been jittered by ±.5 units for clarity. The horizontal and vertical dotted lines represent the median values of the x and y coordinates, respectively.

to drop the dummy variables that indicate whether one or both members of each dyad belong to a free trade agreement, given that such memberships could themselves be dependent on the labor standards of the countries concerned. The model therefore estimates export volumes as a function of

• The product of the distance between the dyad mem- bers

• The product of (logged) GDP of the dyad mem- bers

• The product of (logged) GDP per capita of the dyad members

• A dummy variable for common language

• A dummy variable for a shared land border • A count of the number of landlocked countries in

the dyad • A count of the number of island countries in the dyad • The product of the geographic area of the dyad mem-

bers • A dummy variable for a colonial history with the

same colonial power • A dummy variable for a direct colonial relation-

ship • A dummy variable for situations where both mem-

bers of the dyad were part of the same state • A dummy variable for a common currency • A dummy variable for each year in the series

683

Trade-based Diffusion of Labor Rights November 2009

Presumably, none of the previous variables can be be- lieved to be “caused” by a country’s recent labor rights practices. This allows us to estimate predicted values of bilateral exports that we would expect to find in a hypo- thetical world in which labor rights have no impact on exports. As it turns out, this model manages to account for a large proportion of the variance in exports, with an r square of .61.

In the next stage, we employ the predicted export volumes from this gravity model to recalculate the bi- lateral trade context variables. This produces a measure of the average labor laws (or labor practices) found among each country’s export destinations, weighted by the predicted levels of exports to each destination, rather than the actual volumes of exports. We then reestimate the regression models using these vectors in place of the Bilateral Trade Context: Laws and Bilateral Trade Context: Practices terms. The estimated effects of the key independent variables turn out to be very similar to those reported in Tables 2 and 3: the Bilateral Trade Context: Laws variable is statistically significant at the .05 level in all three of the time-lagged models, and the Bilateral Trade Context: Practices variable is significant at the .05 level after a 3-year lag.21,22 On the basis of these tests, we believe that the possibility of reverse causality does not present a serious problem for the interpretation of our results.

Furthermore, to account for the possibility that labor standards might diffuse among countries as a result of cultural ties (Simmons and Elkins 2004), we include controls for (1) the average labor rights score among countries that share a common language, (2) the av- erage labor rights score among countries that share a common religion, and (3) the average labor rights score among countries that share a common colonial history. Data on ties between countries on the basis of shared languages or religions were obtained from the repli- cation data set of Gartzke and Gleditsch,23 whereas data on previous colonial relationships were obtained from the Central Intelligence Agency’s (CIA’s) World Factbook. We also control for each country’s level of exposure to global cultural norms (or “world society”) by including a measure of the number of IGOs and in- ternational nongovernmental organizations (INGOs) that each state belongs to in each year.24 The inclusion

21 As an additional test, we also re-estimate our vector of predicted levels of exports using a slightly different specification of the Rose model that excludes the GDP per Capita covariate. This addresses the possibility that GDP per Capita might itself be dependent on labor standards, which would threaten the assumption of exogeneity in our predicted levels of exports above. However, this did not result in any significant changes in the estimated effect of the bilateral trade context variables. 22 A summary of the results of these and the other robustness tests discussed in this section are presented in Tables A3 and A4 of the Appendix. 23 In the Gartzke and Gleditsch data set, pairs of countries are con- sidered to share a common language if there is a match between one (or both) of the two most common languages spoken in each country. The same applies for the religion variable. See Gartzke and Gleditsch (2006, 62). 24 Data on the number of IGO and INGO memberships were ob- tained from the Yearbook of International Organizations.

of controls for these alternative pathways of norm dif- fusion has very little effect on the estimated effect of Bi- lateral Trade Context: Laws. Interestingly, we find that the measure of common colonial ties and the INGO variable show a positive and statistically significant re- lationship to Labor Laws, suggesting that cultural ties might also play an important role in the transmission of labor rights norms. However, when these variables are included in our model of Labor Practices, the ef- fect of Bilateral Trade Context: Practices is no longer statistically significant in the 3-year lagged model.

Finally, we account for year-specific effects by in- cluding a series of dummy variables for each year. The Bilateral Trade Context: Laws variable remains robust to this specification, with p values below .01 in all three lagged models. However, the Bilateral Trade Context: Practices variable does not appear to show a statisti- cally significant relationship once the year effects are accounted for.

CONCLUSION Domestic policies are often susceptible to international economic influences. The forces of global economic integration, however, affect labor rights in developing countries in ways that are more nuanced than the pro- ponents or opponents of globalization tend to suggest. Our article provides quantitative evidence that labor standards in developing countries are influenced by the labor standards of their exporting destinations, and not by their overall levels of trade openness. Instead of ex- porters pushing down labor standards of the importing countries as the race to the bottom literature suggests, importers can influence—positively or negatively—the collective labor laws and practices of trade partners. Trade can therefore be considered to provide a conduit for the diffusion of norms and practices regarding labor standards in much the same way that it can be believed to facilitate the diffusion of certain environmental stan- dards.

The finding that a California effect holds with re- spect to collective labor rights has important theoreti- cal and practical implications. Theoretically, it provides further support for the argument that the California effect is not restricted to regulations governing prod- uct standards, but that it can also lead to changes in process standards such as those governing collective labor rights. It also suggests that in the short run, the California effect is more successful at diffusing laws from importing countries to their export partners than it is at changing the actual behavior of firms (or govern- ments) in exporting countries. To a certain extent, this is not surprising, given the frequently observed discon- nect between formal law and actual behavior in many different realms of the social world. However, what this does suggest is that although the California effect provides an explanation for how product standards can diffuse from importing countries to exporting countries based on market forces alone, its ability to transmit process standards from one country to another also depends on the capacity and will of governments in the exporting countries to enforce these standards. In

684

American Political Science Review Vol. 103, No. 4

addition, ongoing compliance with process standards depends on the extent to which information about the exporting countries’ and firms’ level of compliance with the standards can be effectively relayed to stakeholders in the importing countries, and, of course, on the extent to which consumers and stakeholders in these countries care about overseas labor standards.

From a practical point of view, our article suggests ways in which international trading relationships can be managed in order to bring about positive change in the labor standards of an exporting country. Provided that developing countries export most of their goods to countries with higher labor standards than their own, we can expect their labor standards to develop in a more positive direction than would otherwise occur if these countries limited their exposure to global mar- kets. This suggests that labor rights activists ought to re- consider some of their arguments for opposing greater levels of economic integration. Enhancing trade ties, rather than restricting trade flows, may be the best hope for improving labor rights in low- and middle-income nations.

Our analyses also suggest some trade-related issues for future research. Along with examining the question of “with whom” each country trades, future work needs to focus on how the California effect may vary across product categories or industrial sectors. Both trade “in what” and trade “with whom” may have important consequences for the relationship between trade and labor rights. Our speculation is that supply chain pres- sure is likely to vary with the level of brandedness of the traded commodity, a result that is consistent with

case study work by economic sociologists (i.e., Bartley 2005; Gereffi and Korzeniewicz 1994), as well as survey evidence from economists (Elliott and Freeman 2003). Arguably, such pressure might also work quite well in primary products, as has been demonstrated by the backlash in importing countries to the labor practices of Shell in Nigeria. As long as consumers can focus their rewards and sanctions on specific brands (including those in primary sectors such as petroleum), export- ing firms will have incentives to respond to consumer pressure in the importing country. Another possibi- lity is that variation in the skill intensity of production may generate varying incentives for firms to attract and retain workers—leading to interindustry variation in labor rights outcomes and in trade-related diffusion pressures.

Finally, as some developing countries increasingly play a role in global product markets, the direction of the California effect may change. That is, if export destinations with inferior labor rights outcomes be- come more important to global trade generally, and to some countries’ trade specifically, we might find that an average developing country faces less pressure from its trading partners to ratchet up its labor prac- tices. However, if middle-income developing nations improve their labor practices over time, we might see an increased momentum behind trade-induced improve- ments in labor rights.

APPENDIX

TABLE A1. Labor Standards Coding Template

Category Type Description Weight (if Observed)

Freedom of association/collective bargaining–related liberties 1 Practices Murder or disappearance of union members or organizers 2 2 Practices Other violence against union members or organizers 2 3 Practices Arrest, detention, imprisonment, or forced exile for union

membership or activities 2

4 Practices Interference with union rights of assembly, demonstration, free opinion, and free expression

2

5 Practices Seizure or destruction of union premises or property 2

Right to establish and join union and worker organizations 6 Laws General prohibitions 10 7 Practices General absence resulting from socioeconomic breakdown 10 8 Laws Previous authorization requirements. Does not include

requirements that unions register with governments, unless these requirements are deemed onerous by the International Labor Organization (ILO).

1.5

9 Practices Employment conditional on nonmembership in union 1.5 10 Practices Dismissal or suspension for union membership or activities.

Includes dismissal for strike activities. 1.5

11 Practices Interference of employers (attempts to dominate unions) 1.5 12 Practices Dissolution or suspension of union by administrative authority 2 13 Laws Only workers’ committees and labor councils permitted 2 14 Laws Only state-sponsored or other single unions permitted. Includes

allowing only one union per industry or sector. 1.5

15 Laws Exclusion of tradable/industrial sectors from union membership 2

685

Trade-based Diffusion of Labor Rights November 2009

TABLE A1. Continued

Category Type Description Weight (if Observed)

16 Laws Exclusion of other sectors or workers from union membership. Includes exclusion of public sector workers from union membership. Excluding “essential services” is acceptable, provided the definition of “essential services” is not excessively broad (i.e., following ILO guidelines, limitations on armed forces’ union membership are acceptable).

2

17 Practices Other specific de facto problems or acts of prohibition 1.5 18 Laws (No) Right to establish and join federations or confederations of

unions 1.5

19 Laws Previous authorization requirements regarding previous row 1

Other union activities 20 Laws (No) Right to elect representatives in full freedom. Includes

requirement that union leaders must work full time in a given industry.

1.5

21 Laws (No) Right to establish constitutions and rules 1.5 22 Laws General prohibition of union/federation participation in political

activities. Includes limits on union contributions to political parties.

1.5

23 Practices (No) Union control of finances. Includes situations in which unions receive a substantial portion of financing from government sources, or rules that unions may not receive financial contributions from abroad or from certain groups.

1.5

Right to collectively bargain 24 Laws General prohibitions 10 25 Laws Prior approval by authorities of collective agreements 1.5 26 Laws Compulsory binding arbitration. Includes systems in which

compulsory binding arbitration is necessary before a (legal) strike may be called.

1.5

27 Practices Intervention of authorities. Includes unilateral setting of wages by authorities.

1.5

28 Practices Scope of collective bargaining restricted by non–state employers 1.5 29 Laws Exclusion of tradable/industrial sectors from right to collectively

bargain 1.75

30 Laws Exclusion of other sectors or workers from right to collectively bargain. Includes the exclusion of civil servants or all public sector workers. Excluding “essential services” is acceptable, provided the definition of “essential services” is not excessively broad.

1.75

31 Practices Other specific de facto problems or acts of prohibition. Includes “no legal right” to bargain collectively (but no legal prohibition on doing so).

1.5

Right to strike 32 Laws General prohibitions 2 33 Laws Previous authorization required by authorities. Includes

requirement for official approval prior to strike. A requirement to notify officials prior to a strike is not coded as a violation.

1.5

34 Laws Exclusion of tradable/industrial sectors from right to strike 1.5 35 Laws Exclusion of other sectors or workers from right to strike. Includes

the exclusion of civil servants or all public sector workers. Excluding “essential services” is acceptable, provided the definition of “essential services” is not excessively broad.

1.5

36 Practices Other specific de facto problems or acts of prohibition 1.5

Export processing zones 37 Laws Restricted rights in EPZs. Includes EPZs, free trade zones, and/or

special economic zones. 2

Source: Adapted from Kucera (2002). Coding notes from Mosley and Uno (2007) are added in italics.

686

American Political Science Review Vol. 103, No. 4

TABLE A2. Summary Statistics

Variable Mean SD Minimum Maximum

Labor laws 22.85 5.66 .00 28.50 Labor practices 22.46 4.47 .00 27.50 Bilateral trade context: laws 24.67 2.27 10.42 28.43 Bilateral trade context: practices 23.26 2.08 14.48 27.44 FDI inflows 2.75 7.00 −82.87 145.21 Total trade 77.46 44.33 1.53 294.65 GDP per capita (logged) 7.90 .96 5.83 10.21 Population 15.22 2.16 9.55 20.97 Democracy −.40 6.81 −10.00 10.00 Civil war .21 .41 .00 1.00 Hard PTA .02 .15 .00 1.00 Soft PTA .16 .37 .00 1.00

FDI, foreign direct investment; GDP, gross domestic product; PTA, Preferential Trade Agreement.

TABLE A3. Additional Robustness Tests for Labor Laws Model

A1 A2 A3 A4

Bilateral trade context: law .329∗∗ .286∗∗∗

(.111) (.076) Instrumental variable I .286∗

(.111) Instrumental variable II .253∗∗

(.095) Language −.008

(.057) Religion −.082

(.103) Colonial history .235∗∗

(.081) IGOs .029

(.022) INGOs .004∗∗

(.001) Total trade −.001 −.001 −.010 −.003

(.004) (.005) (.007) (.005) FDI inflows .002 .001 .027 .027

(.027) (.028) (.032) (.025) Hard PTA .530 .637 .124 .895

(.616) (.630) (.452) (.471) Soft PTA −.135 −.115 .144 −.064

(.209) (.213) (.308) (.255) GDP per capita −.624∗∗ −.642∗∗ −1.177∗∗ −.452∗

(.213) (.214) (.411) (.184) Democracy .032 .031 .021 .021

(.024) (.024) (.027) (.021) Population −.290∗∗ −.270∗ −1.251∗∗ −.345∗∗

(.106) (.109) (.375) (.105) Civil war −.124 −.103 −.058 −.158

(.244) (.249) (.345) (.235) Lagged dependent variable .649∗∗∗ .643∗∗∗ .544∗∗∗ .622∗∗∗

(.032) (.033) (.064) (.036)

687

Trade-based Diffusion of Labor Rights November 2009

TABLE A3. Continued

A1 A2 A3 A4

Constant 10.138∗∗ 11.062∗∗ 24.371∗∗ 10.165 (3.600) (3.318) (7.541) (2.849)

N 1,222 1,222 691 1,252

IGO, intergovernmental organization; INGO, international nongovermental organization; FDI, foreign direct investment; PTA, Preferential Trade Agreement; GDP, gross domestic product. Notes: All independent variables have been lagged by three years. Models A1 and A2 represent the second stage of the two-stage least squares procedure, where Instrumental Variable I refers to the predicted values of Bilateral Trade Context: Laws obtained using the previously described, model and Instrumental Variable II refers to the modified version of that variable obtained by excluding GDP per Capita (see footnote 21). Model A3 includes controls for alternative paths of norm diffusion. Model A4 includes year-specific dummy variables. The estimates of the individual year effects are not shown. Significance levels are indicated as follows: ∗ p < .05, ∗∗ p < .01, ∗∗∗ p < .001.

TABLE A4. Additional Robustness Tests for Labor Practices Model

A5 A6 A7 A8

Bilateral trade context: .100 .023 practices (.102) (.092)

Instrumental variable I .240∗

(.121) Instrumental variable II .205

(.112) Language −.030

(.075) Religion .163

(.096) Colonial history .103

(.074) IGOs −.005

(.015) INGOs .000

(.001) Total trade .003 .003 −.008 .002

(.004) (.004) (.005) (.004) FDI inflows .007 .004 .008 .016

(.025) (.025) (.035) (.026) Hard PTA .254 .363 .239 .614

(.800) (.804) (.877) (.831) Soft PTA −.380 −.363 −.036 −.148

(.258) (.258) (.293) (.305) GDP per capita −.550∗∗ −.548∗∗ −.435 −.485∗∗

(.164) (.165) (.267) (.162) Democracy .000 .000 −.034 −.010

(.022) (.023) (.027) (.021) Population −.460∗∗∗ −.470∗∗∗ −.825∗∗ −.420∗∗∗

(.124) (.123) (.272) (.116) Civil war −.535 −.514 −1.584∗∗ −.492

(.343) (.345) (.506) (.328) Lagged dependent variable .573∗∗∗ .566∗∗∗ .487∗∗∗ .591∗∗∗

(.030) (.031) (.048) (.029) Constant 14.658∗∗ 15.929∗∗∗ 20.167∗∗ 18.224∗∗∗

(4.495) (3.971) (6.027) (3.698) N 1,222 1,222 691 1,252

IGO, intergovernmental organization; INGO, international nongovernmental organization; FDI, foreign direct investment; PTA, Preferential Trade Agreement; GDP, gross domestic product. Notes: All independent variables have been lagged by three years. Models A5–A8 represent the Labor Practices equivalents of the Labor Laws models presented in Table A3. Significance levels are indicated as follows: ∗ p < .05, ∗∗ p < .01, ∗∗∗ p < .001.

688

American Political Science Review Vol. 103, No. 4

REFERENCES

Abouharb, M. Rodwan, and David Cingranelli. 2007. Human Rights and Structural Adjustment. Cambridge: Cambridge University Press.

Abrami, Regina M. 2003. “Worker Rights and Global Trade: The U.S.-Cambodia Bilateral Textile Trade Agreement.” Harvard Business School Case 703-034.

Aidt, Toke, and Zafiris Tzannaos. 2002. Unions and Collective Bar- gaining: Economic Effects in a Global Environment. Washington, DC: World Bank.

Baron, David. 2003. “Private Politics.” Journal of Economics and Management Strategy 12: 31–66.

Bartley, Tim. 2005. “Corporate Accountability and the Privatiza- tion of Labor Standards: Struggles over Codes of Conduct in the Apparel Industry.” Research in Political Sociology 12: 211–44.

Beck, Nathaniel N., and Katz Jonathan, 2001. “Throwing Out the Baby with the Bath Water: A Comment on Green, Kim, and Yoon.” International Organization 55 (2): 487–95.

Becker, David G., and Richard L. Sklar, eds. 1999. Postimperialism and World Politics. Westport, CT: Praeger.

Blanchflower, David G., and Alex Bryson. 2003. “Changes over Time in Union Relative Wage Effects in the UK and US Revisited.” In International Handbook of Trade Unions, eds. John T. Addison and Claus Schnabel. Northhampton, MA: Edward Elgar, 197–245.

Blanton, Shannon Lindsey, and Robert G. Blanton. 2007. “What Attracts Foreign Investors? An Examination of Human Rights and Foreign Direct Investment.” Journal of Politics 69 (1): 143–55.

Brooks, Sarah M. 2008. Social Protection and the Market in Latin America. Cambridge: Cambridge University Press.

Brune, Nancy, Geoffrey Garrett, and Bruce Kogut. 2004. “The Inter- national Monetary Fund and the Global Spread of Privatization.” IMF Staff Papers 51 (3): 195–219.

Chiu, Catherine C. H. 2007. “Workplace Practices in Hong Kong– Invested Garment Factories in Cambodia.” Journal of Contempo- rary Asia 37 (4): 431–48.

Chweiroth, Jeffrey. 2007. “Neoliberal Economists and Capital Ac- count Liberalization in Emerging Markets.” International Organi- zation 61 (2): 443–63.

Collinsworth, Terry, J., William Goold, and Pharis J. Harvey. 1994. “Labor and Free Trade.” Foreign Affairs 77 (1): 8–13.

Compa, Lance, and Jeffrey S. Vogt. 2001. “The Generalized System of Preferences: A 20-Year Review.” Comparative Law and Policy Journal 22 (2/3): 199–238.

Daude, Christian, Jacqueline Mazza, and Andrew Morrison. 2003. “Core Labor Standards and Foreign Direct Investment in Latin America and the Caribbean: Does Lax Enforcement of Labor Standards Attract Investors?” Interamerican Development Bank. Manuscript.

Elkins, Zachary, Andrew T. Guzman, and Beth A. Simmons. 2006. “Competing for Capital: The Diffusion of Bilateral Investment Treaties, 1960–2000.” International Organization 60 (4): 811–46.

Elliott, Kimberly Ann, and Richard B. Freeman. 2003. Can Labor Standards Improve under Globalization? Washington, DC: Insti- tute for International Economics.

Fair Labor Association (FLA). 2005. 2005 Annual Public Report. Washington, DC: FLA.

Flanagan, Robert J. 2006. Globalization and Labor Conditions: Working Conditions and Worker Rights in a Global Economy. Oxford: Oxford University Press.

Freeman, Richard B. 2007. “Labor Market Institutions around the World.” NBER Working Paper 13242 (July).

Garcia-Johnson, Ronie. 2000. Exporting Environmentalism: U.S. Multinational Chemical Corporations in Brazil and Mexico. Cambridge, MA: MIT Press.

Gartzke, Erik, and Kristian Skrede Gleditsch. 2006. “Identity and Conflict: Ties That Bind and Differences That Divide.” European Journal of International Relations 12 (1): 53–87.

Gereffi Gary, and Miguel Korzeniewicz, eds. 1994. Commodity Chains and Global Capitalism. Westport, CT: Praeger.

Gleditsch, Kristian Skrede, and Michael D. Ward. 2006. “Diffusion and the International Context of Democratization.” International Organization 60: 911–33.

Gourevitch, Peter. 1978. “The Second Image Reversed: The Inter- national Sources of Domestic Politics.” International Organization 32 (4): 881–912.

Greenhill, Brian. Forthcoming. “The Company You Keep: Interna- tional Socialization and the Diffusion of Human Rights Norms.” International Studies Quarterly.

Guthrie, Doug. 2006. China and Globalization. New York: Routledge.

Hafner-Burton, Emilie M. 2005. “Trading Human Rights: How Pref- erential Trade Agreements Influence Government Repression.” International Organization 59: 593–629.

Hafner-Burton, Emilie M., and Kiyoteru Tsutsui. 2005. “Human Rights in a Globalizing World: The Paradox of Empty Promises.” American Journal of Sociology 110 (5): 1373–411.

Hall, Peter A., and David Soskice, eds. 2001. Varieties of Capitalism: The Institutional Foundations of Comparative Advantage. New York: Oxford University Press.

Hathaway, Oona A. 2002. “Do Human Rights Treaties Make a Dif- ference?” The Yale Law Journal 118 (8): 1935–2042.

Huber, Evelyne, and John D. Stephens. 2001. Development and Cri- sis of the Welfare State: Parties and Policies in Global Markets. Chicago: The University of Chicago Press.

Kucera, David. 2002. “Core Labor Standards and Foreign Direct Investment.” International Labour Review 141: 31–69.

Locke, Richard M. 2003. “The Promise and Perils of Globalization: The Case of Nike.” In Management: Inventing and Delivering Its Future, eds. Richard Schmalenese and Thomas A. Kochan. Cambridge, MA: MIT Press, 39–70.

Locke, Richard M., Fei Qin, and Alberto Brause. 2007. “Does Mon- itoring Improve Labor Standards? Lessons from Nike.” Industrial and Labor Relations Review 61 (1): 3–31.

Meseguer, Covadonga. 2004. “What Role for Learning? The Diffu- sion of Privatization in Industrial and Latin American Countries.” Journal of Public Policy 24 (3): 229–325.

Moran Theodore, Edward M. Graham, and Magnus Blomstrom, eds. 2005. Does Foreign Direct Investment Promote Development? Washington, DC: Institute for International Economics.

Mosley, Layna. 2008. “Workers’ Rights in Open Economies: Global Production and Domestic Institutions in the Developing World.” Comparative Political Studies 41: 674–714.

Mosley, Layna, and Saika Uno. 2007. “Racing to the Bottom or Climbing to the Top? Economic Globalization and Collec- tive Labor Rights.” Comparative Political Studies 40 (8): 923– 48.

Neumayer, Eric. 2005. “Do International Human Rights Treaties Im- prove Respect for Human Rights?” Journal of Conflict Resolution 49 (6): 925–53.

Neumayer, Eric, and Indra de Soysa. 2005. “Trade Openness, Foreign Direct Investment and Child Labor.” World Development 33 (1): 43–63.

Neumayer, Eric, and Indra de Soysa. 2006. “Globalization and the Right to Free Association and Collective Bargaining: An Empiri- cal Analysis.” World Development 34 (1): 31–49.

O’Rourke, Dara. 2003. “Outsourcing Regulation: Analyzing Non- Governmental Systems of Labor Standards and Monitoring.” Po- licy Studies Journal 31: 1–29.

Poe, Stephen C., Neal Tate, and Linda Camp Keith. 1999. “Repres- sion of the Human Right to Personal Integrity Revisited: A Global Cross-National Study Covering the Years 1976–1993.” Interna- tional Studies Quarterly 43 (2): 291–313.

Prakash, Aseem. 2000. Greening the Firm. Cambridge: Cambridge University Press.

Prakash, Aseem, and Matthew Potoski. 2006. “Racing to the Bot- tom: Globalization, Environmental Governance, and ISO 14001.” American Journal of Political Science 50 (2): 347–61.

Prakash, Aseem, and Matthew Potoski. 2007. “Investing Up: FDI and the Cross-National Diffusion of ISO 14001.” International Studies Quarterly 51 (3): 723–44.

Quinn, Dennis P. 1997. “The Correlates of Change in Interna- tional Financial Regulation.” American Political Science Review 91 (September): 531–51.

Quinn, Dennis P., and A. Maria Toyoda. 2007. “Ideology and Voter Preferences as Determinants of Financial Globaliza- tion.” American Journal of Political Science 51 (January): 344– 63.

Richards, David L., Ronald D. Gelleny, and David H. Sacko. 2001. “Money with a Mean Streak? Foreign Economic Penetration and Government Respect for Human Rights in Developing Coun- tries.” International Studies Quarterly 45: 219–39.

689

Trade-based Diffusion of Labor Rights November 2009

Rodrı́guez-Garavito, Cesar. 2005. “Global Governance and Labor Rights: Codes of Conduct and Anti-Sweatshop Struggles in Global Apparel Factories in México and Guatemala.” Politics and Society 33 (2): 203–33.

Rose, Andrew. 2004. “Do We Really Know That the WTO Increases Trade?” American Economic Review 94 (1): 98–114.

Siedman, Gay. 2007. Beyond the Boycott. New York: Russsell Sage Foundation.

Simmons Beth A., Frank Dobbin, and Geoffrey Garrett, eds. 2008. The Global Diffusion of Markets and Democracy. Cambridge: Cambridge University Press.

Simmons, Beth A., and Zachary Elkins. 2004. “The Globalization of Liberalization: Policy Diffusion in the International Economy.” American Political Science Review 98 (1): 171–89.

Smith, Gare, and Dan Feldman. 2003. Company Codes of Conduct and International Standards: An Analytical Comparison, Part I of II. Washington, DC: World Bank Group.

Spar, Debora, and L. T. LaMure. 2003. “The Power of Activism.” California Management Review 45 (3): 78–101.

Vogel, David. 1995. Trading Up: Consumer and Environmental Regu- lation in a Global Economy. Cambridge, MA: Harvard University Press.

Vreeland, James Raymond. 2008. “Political Institutions and Human Rights: Why Dictatorships Enter into the United Nations Conven- tion Against Torture.” International Organization 62 (1): 65–101.

Weyland, Kurt. 2007. Bounded Rationality and Policy Diffusion: So- cial Sector Reform in Latin America. Princeton, NJ: Princeton University Press.

690