National Human Trafficking Initiatives: Dimensions of Policy Diffusion

Abstract

The implementation of criminal law involves formal law enforcement, education and public outreach aimed at preventing criminal activity, and providing services for victims. Historically, quantitative research on global trends has tended to focus on a single policy dimension, potentially masking the unique factors that affect the diffusion of each policy dimension independently. Using an ordered-probit model to analyze new human trafficking policy data on national prosecution, prevention, and victim-protection efforts, we find that global ties and domestic interest groups matter more in areas where international law is less defined. While prosecution, officially mandated by the Trafficking Protocol, was relatively impervious to global ties and domestic interest groups, both trafficking prevention and victim protection were associated with these factors. Our findings also suggest that fear of repercussions is not a major driver of state actions to combat trafficking—neither ratification of the Trafficking Protocol nor levels of United States aid were associated with greater implementation of anti-trafficking measures.

INTRODUCTION

Creating change through law is a multifaceted process. Formally adopting a law and enforcing it are obviously key components of this process but, if a state wants the public to recognize and report violations of the law, other actions are important as well. The state must provide education and ideally involve civil society in enforcement efforts. In the case of criminal law, it must also mitigate any negative impacts of reporting so that victims will call attention to violations. In this article, we embrace the complexity of implementing law by considering what independent factors drive countries to take action in each of these areas with a focus on national policies combating human trafficking. We do this by drawing on Cho, Dreher, and Neumayer's (2011a) recently-constructed "3P" Anti-trafficking Policy Index, which measures the extent to which countries 1) criminalize trafficking and prosecute offenders ("Prosecution"); 2) conduct outreach and education to prevent trafficking ("Prevention"); and 3) protect and assist the victims of trafficking ("Protection").

Our analysis of human trafficking considers when and how global ties and local interests affect the three key elements of implementation in different ways. Our core thesis is that less institutionalized and more voluntary aspects of implementation are more sensitive to global and domestic mobilizing. In other words, we expect "world polity" factors and domestic interest groups to matter more in areas where international law is less defined.

Our findings support this expectation. While prosecution, officially mandated by the 2000 Protocol to Prevent, Suppress and Punish Trafficking in Persons, Especially Women and Children (the "Trafficking Protocol"), is relatively impervious to global connections and domestic groups (with the exception of human rights lawyers' organizations), both prevention and protection are associated with these factors. Women's advocacy, in the form of ratification of the Convention on the Elimination of All Forms of Discrimination against Women (CEDAW) and the percentage of women in parliament, appears to play an important role in victim protection. As one might expect, lawyers' professional organizations (LPOs), international nongovernmental organizations (INGOs), women in parliament, CEDAW ratification, and greater democracy, which are all associated with more vibrant civil societies, are positively associated with more education and public outreach (prevention). In addition, our findings suggest that fear of repercussions is not the primary reason why states take actions to combat trafficking—neither ratification of the Trafficking Protocol nor levels of United States aid were associated with greater implementation of anti-trafficking measures.

BACKGROUND

While exact levels of human trafficking are unknown because the practice occurs in the shadow economy, a conservative estimate by the United Nations Office of Drugs and Crime (UNODC 2012) suggests that 2.5 million individuals are exploited by trafficking at any given moment. The UNODC (2009) estimates that about 20 percent of these victims are children. Human trafficking is thus a significant social problem and an important basis for developing theories of legal implementation in the context of globalization.

Article 3(a) of the Trafficking Protocol (2000) defines trafficking in persons as:

the recruitment, transportation, transfer, harbouring or receipt of persons, by means of the threat or use of force or other forms of coercion, of abduction, of fraud, of deception, of the abuse of power or of a position of vulnerability or of the giving or receiving of payments or benefits to achieve the consent of a person having control over another person, for the purpose of exploitation.

From this definition, it is clear that trafficking is distinguishable from smuggling, where the persons transported are (at least theoretically) informed and act voluntarily. In contemporary international law, trafficking is also broader than exploitation for prostitution.² It includes any kind of labor exploitation, such as refusing to pay for work, employing workers illegally in dangerous working conditions, or forcing workers to incur unreasonable debts that are paid off through labor. In fact, the bulk of trafficking victims today are laborers (Feingold 2005).³

Historical Evolution of International Anti-Trafficking Law

International concern over human trafficking has a long history, beginning with the adoption by 13 European countries of the International Agreement for the Suppression of the White Slave Traffic in 1904. This was followed by two League of Nation’s conventions (the Convention for the Suppression of the Traffic in Women and Children in 1921 and the Convention for the Suppression of the Traffic in Women of Full Age in 1933). One of the earliest treaties emanating from the United Nations addressed trafficking as well (the 1949 Convention for the Suppression of the Traffick in Persons and of the Exploitation of the Prostitution of Others).⁴ This 1949 Convention was notable for being the first to criminalize the act of exploiting others for prostitution, although it did not extend to trafficking for other purposes (Gallagher 2010; Lloyd and Simmons 2014). Although it is still technically in effect, the 1949 Convention has been criticized for failing to distinguish between forced and consensual prostitution (Gallagher 2010) and for its lack of protections for victims of trafficking (Coomaraswamy 2000).

Several decades later, in 1979, the Convention on the Elimination of All Forms of Discrimination Against Women (CEDAW) framed the issue of trafficking in terms of women’s human rights. CEDAW required states to take action to suppress "traffic in women and exploitation of prostitution of women" (Article 6). The history of the regulation of trafficking up to and including CEDAW solidified a strong connection between the women’s movement and anti-trafficking mobilization that continues today (Anderson and Andrijasevic 2008).

Nevertheless, in the last three decades, the formal definition of trafficking in international law has steadily expanded (see Ditmore 2005). The 1989 Convention on the Rights of the Child (CRC) broke new ground by forbidding trafficking in children for any purpose, not just sexual exploitation (CRC, Art. 35). An Optional Protocol to the CRC on the Sale of Children, Child Prostitution and Child Pornography, which entered into force in 2002, goes beyond the original convention to adopt an explicitly criminal justice approach to child trafficking (Gallagher 2010).

Other events were also important in redefining trafficking in the 1990s. Early in the decade, the collapse of the Soviet Union and the subsequent war in the former Yugoslavia resulted in a new wave of trafficking from that region (Haynes 2007). At the same time, wealthy states were becoming increasingly concerned that trafficking was circumventing national migration restrictions (Gallagher 2001). Even as lawful opportunities for migration were diminishing, the number of individuals crossing borders was at an all time high (Bhabha 2005). Evidence of organized criminal involvement in trafficking provided affected states with an additional incentive to create a stronger and broader international response (Gallagher 2001; (Lloyd and Simmons 2014).

The Contemporary Anti-Trafficking Regime

These factors were the impetus behind the 2000 United Nations Convention against Transnational Organized Crime (known as the Palermo Convention)⁵ and the related Trafficking Protocol. The Trafficking Protocol radically expanded the definition of trafficking to include all individuals who find themselves in exploitive working conditions as the result of force or deceit, not just women and children. Linked to the Palermo Convention, the Trafficking Protocol solidified a criminalization approach to trafficking (Lloyd and Simmons 2014). At the end of 2011, 147 countries had ratified the Trafficking Protocol.

Broad cross-national support in the battle against human trafficking is consistent with research demonstrating that global institutional environments shape norms and national regulations (Shaffer 2012; Frank, Hardinge, and Wosick-Correa 2009; Berkovitch 1999). This research reveals that global legal trends are most effective and stable within countries when they appeal to different normative orders (Trubek, Mosher, and Rothstein 2000) and consider key interest groups (Halliday and Carruthers 2009). The Protocol does this well. It addresses moral concerns regarding prostitution. It is somewhat responsive to feminist concerns in its calls for protection of the victims of trafficking, as it encourages states to provide trafficking victims with at least temporary residency permits (Trafficking Protocol, Article 7). The Protocol also appeals to labor organizations by providing some victim protections for workers. Furthermore, the broad definition of trafficking in the Protocol satisfies states' interests in targeting illegal immigration (Berman 2010).

In addition to international treaties, individual countries have also passed laws to reduce human trafficking globally. Most notably, the United States’ 2000 Trafficking Victims Protection Act (TVPA) strengthens the Trafficking Protocol by ranking countries' performances in combating trafficking and linking U.S. aid to their reform efforts. The U.S. intentionally minimized differences between the definition of trafficking in the TVPA and that proposed for the Protocol, thereby facilitating international consensus around the concept (Gallagher 2010). Countries identified as lax in the U.S. trafficking reports can lose non-humanitarian and non-trade related aid from the United States and potentially from some international organizations, influencing countries to take trafficking more seriously (Gallagher 2010).

ELEMENTS OF ANTI-TRAFFICKING POLICY IMPLEMENTATION

National efforts to combat trafficking now extend well beyond the US to countries all over the globe. Cho, Dreher, and Neumayer (2011b) report contagion effects regarding compliance with international human trafficking law across countries that are geographically near one another, culturally similar, or, in the case of prosecution, trading partners. Over time, a transnational alliance has emerged to combat human trafficking globally.

As noted earlier, the trafficking framework involves multiple elements. Following Cho et al. (2011a), we identify three here. One step is for the state to place a law on the books and make ideological and resource commitments to enforce it. Another step involves the education of members of the public about the nefarious activity and the recruitment of this public to help reduce it. Finally, the state must protect the victims of criminal behavior from reprisals for coming forward. This protects the human rights of the victims (Coomaraswamy 2000) and, pragmatically, makes it more likely that victims will become whistle-blowers. Without each of these aspects of implementation, states' efforts to combat trafficking, like any other crime, are less likely to be effective.

While research on the global spread of policies tends to focus on the first element of implementation, prosecution, it is important that scholars consider the unique factors that influence these different dimensions of policy implementation (Cho et al. 2011b). Each dimension of implementation may incorporate the interests of particular constituents. For example, it is generally in states’ interests to enforce laws against trafficking because this gives them tighter control over migration and facilitates actions against other types of transnational crime. Protecting the victims of trafficking, in contrast, is less tied to state interests because such protection makes it more difficult to deport trafficking victims (Travaux Préparatoires for the Palermo Convention 2000). State agents are therefore often less enthusiastic in their support of this latter policy dimension, while others, such as feminist-oriented practitioners and organizations, are likely to place greater emphasis on helping victims and less on formal legal actions. Thus, analyzing the components of legal implementation separately allows for a better understanding of the complexity of implementation, which in turn provides clues to the process of how similar laws come to have very different meanings in different countries (Robertson 1995).

We now turn to a discussion of three elements of policy implementation (Cho, Dreher, and Neumayer 2011a).

Prosecution

Prosecution represents the formal legal response to trafficking. The obligation to criminalize intentional trafficking is a core and mandatory provision of the Trafficking Protocol (Gallagher 2010). The prosecution provisions of the Trafficking Protocol in combination with the overall Palermo Convention require states to criminalize the laundering of money gained through trafficking, to extend their statutes of limitations for trafficking crimes, and to take other related steps. In addition, three of the four minimum criteria for state action identified in the United States' TVPA are related to prosecution.

The mandatory nature of these provisions suggests a global consensus, or at least acquiescence, among states around the criminalization of trafficking. Consistent with this, Lloyd, Simmons, and Stewart (2011) found that media stories that mention human trafficking, a specific country, and the words “crime/s” or “victim/s” were strongly predictive of the subsequent criminalization of trafficking in the named country. Because there are both resource incentives and global normative pressures behind prosecution, we expect:

Hypothesis 1A. On average, countries will show greater implementation in prosecution than in prevention or victim protection.

It is important to note that a significant aspect of "prosecution," as this term is used in the trafficking realm, is putting laws on the books that make trafficking a criminal activity. The term does not refer solely or even primarily to the prosecution of criminal cases. At the same time, because bringing criminal cases is one aspect of prosecution, the concept is broader than that used in most previous research on the global diffusion of formal law, which tends to consider only policy adoption (for an important exception, see Frank et al. 2009).

Prevention

Formal legal action does not necessarily result in public awareness of the problem of trafficking or protection for trafficking victims (see, e.g., Bernat and Winkeller 2010). It is thus important to consider which state and global factors influence progress on other dimensions of anti-trafficking policy implementation as well. The next human trafficking policy dimension we consider is prevention, or the level of governmental efforts to prevent and combat human trafficking within civil society. Prevention includes education campaigns, national action plans, information exchange among authorities, and cooperation with nongovernmental organizations and the agencies of other governments (Cho et al. 2011a). Compared to prosecution, prevention captures more informal processes of culture change.

While prosecution is required of states under the Palermo Convention, the prevention provisions of the Convention and related Trafficking Protocol are qualified, making specific state obligations more ambiguous (Gallagher 2010). Article 31 of the Convention calls on states to "endeavor" to develop national projects, establish best practices, reduce opportunities for organized criminal groups to engage in trafficking, and promote public awareness. The Trafficking Protocol has somewhat stronger language, noting in Article 9 that parties to the Protocol "shall establish comprehensive policies, programmes and other measures" to prevent trafficking in persons. Nevertheless, a lack of specificity may create more space for states to avoid engaging in prevention relative to prosecution.

Compared to victim protection, which we discuss in the next section, prevention has certain advantages. Prevention allows states to share the burden of implementation with other states or private actors. For example, exchanging information with others saves states from having to fully investigate every trafficking network by themselves. Further, civil society is a crucial medium through which ideas spread according to world polity theorists (see, e.g., Schofer and Longhofer 2011; Schofer and Hironaka 2005; Boli and Thomas 1999). In addition, several case studies (see e.g., Lebov 2010; Simeunovic-Patic and Copic 2010) have suggested that educating the public (and, in the process, law enforcement personnel) is intimately interconnected with effectively prosecuting trafficking cases and protecting trafficking victims. This may mean that prevention must precede protection. For these reasons, we anticipate:

Hypothesis 1B. On average, countries' levels of prevention will increase more than prosecution over time but less than victim protection.

Protection

The final dimension of policy implementation is protection. This includes government efforts to protect and assist the victims of human trafficking (Cho et al. 2011a:8). Historically, it was assumed that victims of trafficking wanted to return to their countries of origin. This was a convenient assumption by state authorities under pressure to reduce the number of undocumented immigrants in their countries (see, e.g., Lebov 2010). Under this rubric, trafficking victims were typically deported. Increasingly, however, deportation has come to be seen as punishment of the victims.

Nevertheless, the Trafficking Protocol provides little in the way of mandatory actions concerning victim protection (Gallagher 2010). For example, Article 6(3) of the Trafficking Protocol calls on states to "consider implementing" measures to protect the victims of trafficking. While victim advocates claim a wide range of entitlements, including residency permits, housing, medical services, job training, and legal assistance, states have been cautious in embracing victims' rights and the corresponding states obligations (Gallagher 2010).

Victim protection has a substantial cost because it runs counter to many states' migration policies. It also often requires law enforcement officials to rethink and modify their approaches to trafficking victims (Simeunovic-Patic and Copic 2010). We expect the difficulties related to implementing victim protection combined with weak pressure to make state action relatively lax in this sphere. Thus, we hypothesize:

Hypothesis 1C. On average, the level of victim protection will be less than prosecution and prevention.

Consistent with this prediction, Lloyd et al. (2011) found that countries with shared roads tended to adopt similar patterns of trafficking criminalization but not victim protection laws. They surmise that countries feared the displacement of trafficking to their territories if they lagged in the passage of criminal law, but victim protection laws did not generate this same national concern (Lloyd et al. 2011). In sum, we expect those aspects of the international trafficking regime that are less specific and more open to interpretation to elicit lower levels of implementation, with victim protection receiving the least support.

FACTORS AFFECTING EACH ELEMENT OF IMPLEMENTATION

We now consider how different factors may affect each element of implementation. Our focus throughout is how the factors associated with prosecution, prevention and victim protection vary. Overall, we suggest that, because criminalization is taken for granted, less direct advocacy or translation will be needed to insure prosecution. The opposite will be true for dimensions of policy reform where international law is suggestive rather than binding. For these dimensions of policy implementation (prevention and victim protection), links to global civil society and the presence of particular local interests are likely to be influential.

Cho et al's (2011b) own analysis of the 3P data hones in on who emulates whom in the adoption of anti-trafficking measures and provides an important foundation for this work. Our work moves beyond their important study to explicitly theorize and test differences in the factors that influence prosecution, prevention, and protection. Importantly, we consider world polity factors, such as links to international organizations, as well as the role of lowers in the process of implementation. In this section, we first consider factors related to the global anti-trafficking regime in particular, then we focus on world polity society factors, and finally we turn our attention to national characteristics.

Prior anti-trafficking research has shown that the adoption of reforms does not occur independently, but rather is a global process (Cho et al. 2011b; Lloyd et al. 2011). Related work in the world polity tradition indicates that, once an idea is established in the international system, it tends to be quickly incorporated into the national policy of many countries in the world (Meyer 2004). This work illustrates the interconnection among states and between states and other global actors (Boli and Thomas 1999).

The terms of the Trafficking Protocol plus its connection to an organized crime treaty suggest that, among states, there is consensus or at least acquiescence around a criminalization approach that privileges prosecution (see Lloyd and Simmons 2014). We therefore expect that countries ratifying the Trafficking Protocol will be especially committed to prosecution:

Hypothesis 2A: Countries that have ratified the Trafficking Protocol will on average have greater levels of prosecution than those that have not.

In contrast to mandating prosecution, the Trafficking Protocol merely recommends action concerning prevention and victim protection, leading to an additional hypothesis:

Hypothesis 2B: Ratification of the Trafficking Protocol will be more strongly associated with prosecution than with prevention or victim protection.

In addition to the Trafficking Protocol, through the TVPA, the United States acts as a "global sheriff" in the realm of human trafficking (Chuang 2006). It provides carrots and sticks to countries to take particular actions, including linking foreign aid to reform (Merry 2014). The TVPA establishes a three-tier system; countries falling to the third tier face sanctions. Case studies indicate that the tier system prompts action. To provide just one example, Cambodian police officers staged a major raid on a brothel in 2002 after the country was placed on the United States' Tier 3 list (Cotter 2008). The minimum criteria for state action in the TVPA focuses heavily on prosecution. We therefore also expect U.S. aid to a country to be significantly and most strongly associated with prosecution:

Hypothesis 3A: The more foreign aid from the United States that a country receives, the greater that country's implementation of prosecution.

Hypothesis 3B: US aid will be more strongly associated with prosecution than with prevention or victim protection.

To summarize, we anticipate states that are bound by law or resources to the global anti-trafficking regime will prosecute trafficking more strongly than other states. We also expect links to the global anti-trafficking regime to be more influential for prosecution than for prevention or victim protection.

Beyond the formal international anti-trafficking regime, research in the world polity tradition reveals that countries that are more embedded in the global system are more likely to adopt policies and to do so earlier than other countries (Meyer et al. 1997; Frank, Hironaka, and Schofer 2000). For example, Meyer et al. (1997) showed that countries that had greater links to international nongovernmental associations (INGOs) were more likely to adopt environmental reforms. For Meyer et al., INGOs signify embeddedness in global civil society. Merry's (2006) global ethnography also notes the significance of INGOs but suggests a different process at work. She pinpoints the necessity of liminal actors who can translate new ideas into the everyday vernacular of communities. INGO activists can be a particularly important source of this type of translation. Keck and Sikkink (1998) also found that INGOs were critically important in pressuring states to conform to the terms of treaties they had ratified.

We anticipate that this relationship will be strongest for prevention because INGOs are part of civil society and prevention is the element of implementation most closely linked to civil society.

Hypothesis 4A: The level of INGOs within countries will be positively associated most strongly with prevention.

Consistent with our overall prediction that world polity factors will matter more for the soft law elements of the anti-trafficking regime, we also hypothesize:

Hypothesis 4B. The level of INGOs within countries will be more strongly associated with protection and prevention than with prosecution.

Another measure of countries' embeddedness in the world polity particularly relevant to human trafficking is their ratification of CEDAW. While the Trafficking Protocol mandates that states take action to make trafficking a crime, it is less demanding with respect to prosecution and victim protection. Consequently, we expect it to have less impact on state performance in these areas. We expect CEDAW to operate differently. With its focus on women's rights, this convention should have the greatest impact on victim protection. For example, Yoo (2011) found that the ratification of CEDAW was associated with greater respect for women’s social rights within states.⁶ Although the definition of trafficking expanded in the late 20th Century to include any gender, we still anticipate a connection between concerns for women's rights and concern for trafficking victims.

Hypothesis 5A: Ratification of CEDAW will be positively associated most strongly with victim protection.

In addition, because of its close connection with women's rights organizations, we expect CEDAW ratification to be associated with the soft law elements of the anti-trafficking regime:

Hypothesis 5B. Ratification of CEDAW will be more strongly associated with prevention and protection than with prosecution.

In moving beyond the global diffusion of laws to consider how the implementation of laws unfolds across levels, other work identifies the importance of local actors. Halliday and Carruthers (2009) suggest that the involvement of particular interest groups is critical for global reforms to be successfully implemented at the national level. If key groups are omitted from the implementation process, "actor mismatch" occurs, decreasing the chances of stable and effective legal reform. While Halliday and Carruthers focus on the extent to which policies are implemented in general, it makes sense that domestic groups will also drive which aspects of policies are more or less implemented. This notion is supported by Cho et al. (2011b), who found, for example, that the percentage of women in parliament is associated with all three aspects of implementation.

Consistent with Cho et al.'s (2011b) findings, we anticipate first that, where women are well represented, feminist concerns will dominate, and anti-trafficking efforts will be more fully embraced. This should lead especially to improved implementation of victim protection. As a comparison case, Frank et al. (2009) found that the presence of women's associations within a country was related to greater levels of police reporting of rapes.

Hypothesis 6A: The greater the percentage of legislators within a country who are women, the greater the level of victim protection in that country.

Extending our argument that local interest groups will have the greatest effect on the soft law elements of trafficking, we expect the presence of women in parliament to be especially influential for victim protection and prevention:

Hypothesis 6B: Greater percentages of legislators within a country who are women will be more strongly associated with prevention and protection than with prosecution.

Lawyers can also constitute an interest group in terms of encouraging the use of the legal system and hence prosecutions. Munger (2015) found that cause lawyers in Thailand were able to increase prosecutions against traffickers. Their ability to affect victim protection was more constrained because of the system's emphasis on criminalization. For a global problem such as trafficking, the presence of lawyers with a global human rights orientation seems particularly relevant. We expect the presence of international lawyers professional organizations (LPOs) that are oriented toward promoting human rights within countries to result in an emphasis on formal law concerning trafficking:

Hypothesis 7: The greater the number of lawyer's international professional organizations within a country, the greater the level of prosecution.

In general, women's empowerment, more lawyers, and more INGOs within a country suggest a vibrant civil society ready to engage the issue of trafficking and partner with the state to reduce the problem (see Berkovitch 1999; Halliday and Karpik 2012). Within countries, the particular constellation of empowered interest groups will affect the nature of implementation.

Finally, previous research has also shown that global power differences affect countries' participation in global trends (see, e.g., Barrett et al. 2010; Halliday and Carruthers 2009). In the case of human trafficking, Eastern European countries during the time of our analysis were under particular pressure to comply with the Palermo Convention because of their desire to join the European Union (EU), as legislative and administrative processes related to trafficking were prerequisites to EU membership for these candidate countries (Haynes 2004: 224–225). Empirically demonstrating the impact of this pressure, the UNODC (2009) identified increases in the number of trafficking convictions in Eastern Europe and Central Asia from 2003 to 2007. Because the formal terms of the anti-trafficking regime emphasize criminalization, we expect these countries to be under the greatest pressure to implement prosecution:

Hypothesis 8. Eastern European countries will on average show higher levels of prosecution of human trafficking than countries in other regions of the world.

Interestingly, eagerness to join the EU has likely diminished since the data for this study were made available; therefore future research might show less distinctiveness for the East European region.

We include a measure of national affluence in our models as a control. National wealth is a marker of state capacity (see Lebov 2010). Wealth will also tend to distinguish countries where trafficking originates from countries that are trafficking destinations. Poverty is often regarded as the root cause of trafficking, but the linkages between poverty, a lack of development and trafficking are complex. For example, there is some evidence to suggest that victims of cross-border trafficking are more likely to originate from middle-income rather than lower-income countries (Danailova-Trainor and Laczko 2010). Even with this complexity, national wealth becomes an important control variable in our models.

In summary, we anticipate that prosecution will be the most fully implemented aspect of the Trafficking Protocol because the prosecution provisions of the Protocol tend to be mandatory, signaling both more enforcement potential and more global consensus around this sphere of implementation. We expect links to the world polity and domestic political interest groups to be more influential in urging countries to adopt prevention and victim protection, while the ratification of the Trafficking Protocol itself and the presence of U.S. foreign aid will influence levels of prosecution within countries. A failure to ratify the Trafficking Protocol may signify a rejection of even minimal efforts to make trafficking a crime, and countries with high levels of aid may fear losing those funds if they fail to criminalize trafficking. Finally, the unique history of Eastern European countries may set them apart from the rest of the world in terms of criminalizing trafficking.

DATA, MEASUREMENT AND ANALYTICAL PLAN

Our dataset examining the effects of global and national characteristics on anti-human trafficking efforts comprises 105 countries from 2000 to 2008. For each dependent variable, the number of observations is 551 (prevention), 555 (protection) and 557 (prosecution), respectively. The list of countries included in the analysis is shown in Appendix A. The dataset has 28 OECD member countries (signifying countries with larger economies) and 72 non-OECD member countries. Table 1 shows the descriptive statistics for the variables included in our models.

Table 1.

Descriptive Statistics for Trafficking Index Outcomes, National Characteristics, and World Polity Measures.

Variables	Mean	Standard Deviation	Min	Max
Dependent Variables
Prevention	2.344	0.908	0	4
Protection	1.947	1.056	0	4
Prosecution	2.834	1.163	0	4
National Characteristics
Level of Democracy	4.994	5.66	−9	10
GDP per capita (logged)	8.671	1.287	5.518	10.88
Trade Openness (% GDP)	82.163	48.457	20.485	438.092
Human Rights Practice	2.223	1.003	1	5
Women's Percentage in National Parliaments	15.755	9.617	0	49
Eastern Europe (dummy)	0.249	0.433	0	1
World Polity Variables
Trafficking Protocol Ratification (dummy)	0.568	0.496	0	1
CEDAW Ratification (dummy)	0.978	0.145	0	1
INGO membership (/100)	12.66	9.98	1.25	41.31
INGO membership (logged)	6.82	0.838	4.836	8.327
Legal Professional Organizations	6.965	9.735	0	73
Lagged Dependent Variables
Prevention	2.211	0.938	0	4
Protection	1.837	1.056	0	4
Prosecution	2.671	1.192	0	4

Dependent Variables

Cho et al.'s (2011a) 3P Anti-trafficking Policy Index measures anti-trafficking policies in 177 countries from 2000 to 2009. These researchers evaluated government compliance with the three elements of the Trafficking Protocol, including legislative adoption, enforcement, implementation and cooperation to "punish and prosecute traffickers," "prevent and combat trafficking in persons, paying particular attention to women and children," and to "protect and assist the victims of such trafficking, with full respect for their internationally recognized human rights" (Cho et al. 2011a:1). Their data are based in part on the Annual Reports of Trafficking in Persons (United States Department of State 2001–2010), but they also draw information from the Global Reports on Trafficking in Persons (UNODC 2006–2009). Following the definition of trafficking in persons adopted by the Trafficking Protocol, the index scores the three dimensions of anti-trafficking policies on scales from 1 to 5. To aid in our statistical analysis, we recoded the categories to range from 0 to 4 and merged categories 0 and 1 into one category. The higher the score, the better the practices existing in that country for that year.

Prosecution of Traffickers

The prosecution index focuses on 1) whether a country has legislative and other means to establish criminal offenses for trafficking in persons; and 2) whether such legislative and other measures are effectively enforced. The level of each government's practices to prosecute trafficking offenders is determined based on whether the country has adopted an anti-trafficking law, a child trafficking law, and other relevant laws. In addition, the index considers the stringency of penalties, the level of law enforcement, and the collection of crime statistics regarding trafficking (Cho et al. 2011a). Notably, only the highest score of 4 signals actual implementation of laws on the books. Scores of either 0 or 1 indicate that a country has no specific laws concerning trafficking.

Prevention of Trafficking

The prevention index measures whether a country establishes and practices comprehensive policies, programs and other measures to prevent trafficking in persons (Cho et al. 2011a). Specifically, the index measures the governments' anti-trafficking campaigns and national action plans, training of government officials, facilitation of information exchange among authorities, monitoring borders, and cooperation with INGOs and other governments. The level of implementation varies as follows: very strong (4), strong (3), modest (2), limited (1) or none (0). The ranking focuses more on the nature of the policies and programs than on their actual implementation.

Protection of Victims

The protection index measures whether a country provides legal protections to victims of trafficking. The index was constructed based on state actions encouraged by the Trafficking Protocol (article 6–8; Cho et al. 2011a). Specifically, it measures whether a government punishes victims, imposes no self-identification to prove their victim status, assists the victim in legal proceedings and repatriation, and provides residence permits and basic services, including housing, medical training, job training, and rehabilitation. The Cho et al. protection scale focuses more on the existence of programs and policy statements rather than the actual level of investment of staff and other resources by countries.

Overall, countries vary widely on these indices. In 2008, the following countries had the highest possible score on each of the three indexes: Belgium, Denmark, Germany, Italy, South Korea, The Netherlands, Norway, Sweden, and the United States. Only one country received the lowest possible score on each dimension in 2008: North Korea. From 2000 to 2007, countries that received the lowest possible cumulative score at least once included Afghanistan, Cuba, Djibouti, Guyana, Iraq, Lebanon, Saudi Arabia, Sudan, Somalia, and Turkmenistan.

Independent Variables

International Organizations

To assess the effects of national linkage to the world society, we measure the number of nation's memberships in international organizations from the Yearbook of International Organizations (Union of International Associations various years). We include both governmental and non-governmental organizations although some prior research has considered only nongovernmental organizations. We do this because international governmental organizations play a particularly important role in regulating human trafficking and coordinating activities across countries in the trafficking area. We focus on international organizations rather than domestic organizations because the former are a better indicator of links to global norms. In addition, prior research has found that global factors are the catalyst for domestic NGOs in many countries (Longhofer and Schofer 2010). These data measure country-level memberships annually from 2000 to 2008. The number is logged to correct for skew.

Treaty Ratifications

The two treaties most relevant to trafficking are CEDAW and the Trafficking Protocol. Dummy variables capture each state's ratification status in each year. By 2012, 187 countries had ratified CEDAW and 147 countries had ratified the Trafficking Protocol.

United States Foreign Aid

The United States provides aid for the promotion of human rights, which, by its accounting, includes aid to combat trafficking. Data on levels of U.S. aid for human rights for each country come from AidData 2.0 (Tierney et al. 2011), which does not disaggregate aid within the general category of human rights. The aggregate human-rights-aid measure is appropriate at any rate because all non-trade, non-humanitarian aid from the U.S. is linked to compliance with the TVPA. In other words, if a country fails to meet the minimum TVPA requirements, all of its human rights funding from the U.S. is at risk. We use actual disbursements rather than future commitments for aid. We anticipate that countries that receive more aid will be the most responsive to pressure from the TVPA.

Women's Percentage in National Parliaments

We employ the percentage of women in national parliaments as an indicator to measure women's power in a given country. This measure is collected from World Development Indicators (World Bank 2009; see also Paxton, Hughes, and Green 2006).

Legal Professional Organizations (LPOs)

We also use data measuring the number of legal professional organizations promoting human rights within a given country. The data are collected from the International Rule of Law Directory (International Bar Association 2013). The database keeps records of worldwide legal organizations that promote human rights. Specifically, the directory collects data on organizations offering assistance in the administration of justice; offering assistance to arbitrators, mediators and conciliators in developing access to justice; monitoring, advising and assisting lawyers, human rights defense and police/prison workers; offering assistance to law schools and universities, promoting the legal protection of human rights; offering assistance to refugees, asylum seekers, migrants and internally displaced people; offering financial assistance to uphold the rule of law; making recommendations and promoting law reform that guarantee fundamental human rights and civil liberties; and offering publication of research findings, public lectures, regular conferences and seminars in the rule of law field.⁷

We record the founding year of LPOs in a host country and cumulate the number of LPOs annually. The cumulative numbers of LPOs predicts how active LPOs are in promoting human rights in a given country.

Eastern Europe

This is a dummy variable that captures the former Soviet Union. The Eastern Europe dummy variable follows the regional classification of the World Bank.

Level of Democracy

To examine whether the level of democracy in a country affects the level of anti-human trafficking policy, we use the policy score from Polity IV dataset (Marshall, Jaggers, and Gurr 2010). A score of 10 suggests a full democracy while a score of −10 indicates a full autocracy.

Level of Economic Development

Countries with larger economies may have more capacity to implement trafficking reforms. We therefore include the World Bank's (2009) measure of gross domestic product (GDP) per capita. We logged the indicator to avoid skew. We also included membership in the Organization of Economically Developed Countries (OECD) as another way of distinguishing advanced liberal democracies from other countries. (Although the OECD is not limited to liberal democracies, these countries are overwhelmingly represented among the membership).

Control Variables

Trade Openness

Previous analysis has shown that countries learn human rights practices from their peer trading partners (Cho et al. 2011a). In addition, prior research has shown that countries that are more open to trade provide better economic rights to women and have a lower incidence of forced labor (Neumayer and de Soysa 2007). Thus, we include trade openness as a control variable and use a measure of trade as a percentage of GDP from the World Bank (2009).

Lagged Dependent Variable

To control for unobserved heterogeneity in panel data, we employ one-year lagged dependent variables (Cho et al. 2011b; Wooldridge 2011). These indicate the effects of past level of anti-human trafficking implementation on the current level of anti-human trafficking implementation (Greene 2010). The inclusion of this variable also focuses the analysis on change rather than on initial differences among countries.

Other Control Variables

We included other control variables that were ultimately omitted from the analysis reported here because they did not significantly change the models. To test for potential differences between "sending" and "receiving" countries, we controlled for population and levels of remittance inflows as a percentage of GDP. To test the stability of our models in light of controls included in earlier analyses (Cho et al. 2011b), we included a contiguity-weighted spatial lag and measures of women's economic rights and country corruption. We report our findings concerning these variables along with our other results.

Analysis

To take account of the ordered nature of the dependent variables, this analysis uses an ordered probit model (Wooldridge 2011; Greene 2010). Ordered probit analysis allows researchers to interpret the results as in OLS regression. The model assumes unequal differences in the ranking order of dependent variables (McKelvey and Zavoina 1975; Greene 2010). This feature helps to capture the qualitative aspect of Cho et al.'s trafficking policy indexes. Following Greene (2010), the model is:

$${y_{it}}^{*}={\alpha }_t+\beta \prime y_{it-1}+\beta ″x_{it}+{\epsilon }_{it},$$

Here, y_it is a trafficking index score of state i at the time of t. β′ and β″ are estimates of independent variables, and α_t is the intercept term.

We used methodological checks to verify the robustness of our findings. The logic of ordered probit models is similar to traditional regression models.⁸ However, the model can be refuted if the effect of coefficients is not constant over time. To address this issue, we ran time fixed effects models for each dependent variable with time-varying independent variables (analysis available on request). The results demonstrated that the effects of our key independent variables did not substantially deviate from our reported results. We report the one-level model here so we can show the effects of non-time-varying independent variables, such as being part of Eastern Europe.

We also tested the robustness of our analysis by replacing all time-varying variables with one-year lagged variables. A two-stage least squares (2SLS) model using instrumental variables for CEDAW ratification and Legal Professional Organizations alleviated concerns about possible endogeneity (results are not shown because the IV strategies report only test statistics). Finally, we tested the marginal effects of the ordered probit analysis to ensure that the effects of the independent variables were the same across all dependent variable categories (analysis available on request). We conclude that our analysis is robust against proportional odds assumptions and against within-country variations.

RESULTS

Figure 1 shows the general trends in prevention, protection, and prosecution across all countries in our sample (N=105). Again, each dimension can range from zero to four. From 2000 to 2002, there is an uptick in the average of all three types of implementation. This may reflect countries' execution of the basic requirements of the Palermo Convention. Figure 1 reveals that prosecution, which is mandatory under the Trafficking Protocol and represents formal compliance with global norms, increases the most during this period, providing support for Hypothesis 1A. Countries on average scored higher on prosecution than on prevention or protection throughout the period (with the exception of 2002), and this dimension showed the overall greatest increase over time. Notably, only the highest score on the index scale signifies actual criminal prosecutions taking place within countries. The global average was a full point below the highest possible score, indicating that there is still much room for improvement, even on the most strongly implemented dimension of trafficking policies. Prosecution in this context is mostly a signal that countries are criminalizing trafficking on their books.

Figure 1.

Mean values for Prosecution, Prevention, and Protection Index Scores over time (Range: 0–4)

Supporting Hypothesis 1C, protection is the least implemented dimension of the anti-trafficking policies. This is consistent with the conflation of trafficking with migration policy and the associated political and financial costs of protection. The Trafficking Protocol does not have strong requirements concerning protection, suggesting that there is less state consensus around this dimension of the global effort to combat trafficking. On the positive side, protection did show some increase over time on average. Indeed, Gallagher (2010) suggests that countries are beginning to realize that protection and prevention are necessary foundations for successful prosecution.

Turning next to our ordered probit analysis, we consider the impact of national and international factors on each of the three dimensions of policy implementation from 2000 to 2008. Models 1 and 2 in Table 2 relate to prosecution, Models 3 and 4 relate to prevention, and Models 5 and 6 relate to victim protection. All of the models include the same 105 countries.⁹ International organizations and legal professional organizations were highly correlated so we do not include them together in the same models.

Table 2.

Ordered Probit Analyses of the Impact of National and International Factors on Levels of Prosecution, Prevention, and Victim Protection concerning Human Trafficking, 2000–2008.

	(1)	(2)	(3)	(4)	(5)	(6)
	Prosecution		Prevention		Protection
Trafficking Protocol	−0.061 (0.118)	−0.047 (0.118)	0.001 (0.102)	0.006 (0.106)	0.051 (0.100)	0.040 (0.096)
U.S. Aid (Constant US$, logged)	0.013 (0.012)	0.013 (0.012)	−0.004 (0.010)	−0.005 (0.010)	−0.006 (0.009)	−0.008 (0.009)
Int'l Orgs (logged)	0.069 (0.114)		0.192* (0.097)		0.214* (0.101)
CEDAW	0.302 (0.420)	0.251 (0.415)	1.017*** (0.258)	1.012*** (0.267)	0.640+ (0.346)	0.680* (0.319)
Percent Women in National Politics	0.003 (0.006)	0.003 (0.006)	0.014* (0.006)	0.016** (0.006)	0.013* (0.006)	0.014* (0.006)
Legal Professional Orgs		0.016** (0.006)		0.010* (0.004)		0.003 (0.002)
Eastern Europe (dummy)	0.478** (0.160)	0.473*** (0.138)	0.172 (0.126)	0.113 (0.115)	0.079 (0.138)	0.001 (0.122)
Level of Democracy	0.014 (0.013)	0.013 (0.013)	0.019+ (0.011)	0.022* (0.011)	0.027** (0.010)	0.031*** (0.009)
GDP per capita(logged)	0.163* (0.078)	0.172** (0.056)	0.112 (0.071)	0.183** (0.056)	0.053 (0.076)	0.146* (0.057)
Trade Openness	−0.000 (0.001)	0.000 (0.001)	−0.001 (0.001)	−0.001 (0.001)	−0.000 (0.001)	−0.001 (0.001)
Lagged DV(t−1)	1.269*** (0.071)	1.266*** (0.072)	0.885*** (0.087)	0.876*** (0.090)	0.943*** (0.085)	0.948*** (0.085)
cut1 _cons	0.540 (0.668)	0.620 (0.614)	2.401*** (0.499)	1.891*** (0.435)	2.419*** (0.575)	1.841*** (0.525)
cut2 _cons	3.918*** (0.659)	3.977*** (0.604)	3.938*** (0.555)	3.429*** (0.490)	3.930*** (0.569)	3.346*** (0.516)
cut3 _cons	4.702*** (0.679)	4.760*** (0.627)	5.736*** (0.600)	5.218*** (0.523)	5.351*** (0.580)	4.762*** (0.532)
cut4 _cons	6.521*** (0.720)	6.592*** (0.665)	7.423*** (0.659)	6.922*** (0.571)	7.084*** (0.622)	6.487*** (0.569)
N	557	557	551	551	555	555
ll	−421.314	−419.282	−520.842	−518.981	−571.761	−572.820

Standard errors in parentheses (two-tailed tests)

⁺p < 0.10,

^*p < 0.05,

^**p < 0.01,

^***p < 0.001

Differences are notable in the associations between the independent variables and each of the dimensions of trafficking policy implementation. While levels of prevention and protection are associated with mostly the same cluster of indicators, levels of prosecution appear to operate quite differently. This is consistent with our contention that dimensions of implementation follow unique trajectories.

We hypothesized that countries that have ratified the Trafficking Protocol would show more improvement on prosecution than countries that have not because prosecution is mandatory under the Trafficking Protocol. Surprisingly, our ordered probit analysis shows no impact of Trafficking Protocol ratification on prosecution, all else being equal. Thus, Hypotheses 2A is not supported. This may be a consequence of the relatively recent introduction of the Palermo Convention. Ratifying the Trafficking Protocol had no statistically significant effect on prevention or victim protection. This is not surprising, given the soft approach of the Protocol toward these dimensions of trafficking policy (Hypothesis 2B).

The level of bilateral aid from the United States to support human rights within countries also had no statistically significant impact on any of the three dimensions, although the coefficients are in the directions predicted by Hypotheses 3A and 3B. This confirms earlier research findings (Cho et al. 2011b), and could mean that the U.S. is not obviously favoring its allies in its reports. It may also suggest that U.S. resources are not driving national policies on trafficking. Because the U.S. does not give aid to wealthy countries, it is also possible that the GDP per capita effect is cancelling any effect related to aid. To test this, we included an interaction term for GDP per capita and U.S. human rights-related aid in the models (not shown), but it had no significant effect on any of the outcomes. Another possibility is that fear of losing aid brings states to the minimum requirements but nothing beyond that.

The findings of our quantitative analysis also support our contention that less institutionalized and more voluntary aspects of implementation are more sensitive to global and domestic influences. With respect to the indicators representing world polity factors, we find a positive association between international organizations and prevention and protection, but no effect of international organizations on prosecution, supporting Hypothesis 4B. Our findings are consistent with the prediction that INGOs operate to mobilize civil society, although the association of international organizations with prevention and protection is in fact roughly comparable (not supporting Hypothesis 4A). They are also consistent with existing research. For example, the association of INGOs with prevention is consistent with the finding of Frank et al. (2009) that the presence of INGOs strengthens the link between new rape laws and police reporting of rapes. Our finding suggests that INGOs facilitate the education of the public and law enforcement about trafficking.¹⁰ The INGO indicator link to protection is consistent with an overall global trend toward concern for individuals and their rights (Frank, Camp, and Boutcher 2010). It is also consistent with our core claim that world polity factors matter more in areas where international law is less definitive.

Previous research suggests a strong connection between international organizations and the diffusion of formal laws. Our finding that INGOs are not significantly associated with prosecution implies that world polity ties matter less for mandatory aspects of international law. It appears that links to the world polity through international organizations work more strongly on less formal policy dimensions. This makes sense because, in these latter areas, shaming may matter more (Keck and Sikkink 1998). The INGO effect was no longer statistically significant in models (not shown) in which we controlled for OECD membership. Thus, an alternative explanation of the INGO association is that wealthier countries, which tend to have more INGOs, also tend to go beyond prosecution to implement prevention and protection measures.

In addition to the presence of international organizations, countries' ratification of international treaties is a sign of their embeddedness in the global community. In contrast to the Trafficking Protocol, ratification of CEDAW is significantly and positively associated with prevention (Models 3 and 4) and victim protection (Models 5 and 6), supporting Hypothesis 5B. Ratification is most strongly associated with prevention rather than protection, refuting Hypothesis 5A. Apparently, actions that are initially mostly symbolic, such as the ratification of human rights treaties, allow experts and activists to gain leverage in pressuring states for reform (Keck and Sikkink 1998). It is also important to note that CEDAW is 20 years older than the Palermo Convention and so has had more time to develop an effective monitoring system.

We find no associations between prosecution and any of our measures of global pressure or embeddedness in global civil society. This suggests that a developing consensus around prosecution is moving states to take a strong stand on this aspect of combating trafficking independent of their global connections. It is also possible that states' commitment to prosecution leads international actors, such as an INGO activists and CEDAW Committee members, to focus their attention on other aspects of implementation.

Turning to the impact of domestic characteristics, we first consider the percentage of national political representatives who are women. The percentage of women in a country's national parliament had no significant association with the prosecution of human trafficking laws but had significant and roughly equal positive connections to prevention and protection (Hypotheses 6A and 6B). The finding suggests, especially when considered in combination with the impact of ratification of CEDAW, that female legislators have been active in developing systems to educate the public about trafficking (Models 3 and 4). Furthermore, the results may reflect an increasing awareness that women's social and economic circumstances are as important as force and deception in explaining why women get trafficked. In keeping with this, women legislators promote the protection of the victims of trafficking (Models 5 and 6).

Our analysis reveals that women as an interest group may be more concerned with civil society mobilization and victim protection than with the adoption and enforcement of formal laws. Or they may devote their resources to ensuring protection and prevention, assuming that prosecution will develop without their help. That the ratification of CEDAW and women in parliament predict prevention and protection (but not prosecution), while ratification of the Trafficking Protocol predicts nothing, suggests that women's advocacy has had a critical impact on the international anti-trafficking regime and that this impact operates apart from rather than as a reinforcement of formal legal provisions.

International legal professional organizations (LPOs) representing the domestic presence of human rights lawyers are somewhat correlated with international organization links; thus, the two indicators are considered separately in Table 2. Model 2 reveals a modest positive association between lawyers' organizations and the prosecution of human trafficking, supporting Hypothesis 7. As expected, a more organized and globally-focused bar has no significant effect on the protection of trafficking victims, however. Model 4 in Table 2 shows that human-rights-lawyers' organizations are associated with greater prevention as well. Overall, our results support the generalizability of the findings of an in-depth case study of cause lawyering in Thailand (Munger 2015) that anti-trafficking lawyers were ultimately more committed to successful prosecution than to sheltering those who had been trafficked.

Because this finding has the potential to connect the global diffusion literature with the cause lawyering literature, we elaborate on it here. The statistically significant positive association between LPOs and prevention may indicate links between lawyers and civil society actors (see, e.g., Bonelli 2003; Halliday and Karpik 2012). Our results suggest that when the number of lawyers’ international organizations is high within a country, lawyers are more likely to facilitate civil society support for human trafficking laws.

To create a more thorough illustration, we provide the expected probabilities for LPOs at the indicator’s minimum and maximum values, using the software program CLARIFY (Tomz, Wittenberg, and King 2003). Figure 2 shows the expected probabilities for the prevention policy index. If a country has no LPOs, its probability of being in the highest category of 5 (on a scale of 1 to 5) for prevention is 2.5% while the probability of a country with the maximum number of LPOs (n=334) being in this category is 74.4%. On average, the expected probabilities of being in the lower categories of the prevention policy index are much higher in a country without LPOs. Figure 3 also reveals a similar distribution for the prosecution policy index. The expected probability of being in the highest category of 5 is 95% when the number of LPOs is at its peak. In combination, the positive associations of INGOs, CEDAW ratification, women in parliament, and LPOs suggest that countries with vibrant civil society are stronger on prevention than other countries.

Figure 2.

Prevention, Expected Probabilities of Receiving a Particular Index Score if Legal Professional Organizations Is at Its Minimum or Maximum Values.

Figure 3.

Protection, Expected Probabilities of Receiving a Particular Index Score if Legal Professional Organizations Is at Its Minimum or Maximum Values.

Eastern European countries scored significantly higher on prosecution compared to all other regions of the world, when other factors were controlled (Models 1 and 2). This provides support for Hypothesis 8. During the period studied, many of these countries were interested in being integrated into the European Union and therefore had a strong incentive to show control over their borders. On the Cho et al. (2011a) five-point scale, only the top score indicates actual implementation of laws on the books, so this finding does not necessarily imply that Eastern European countries are bringing many more trafficking cases than other countries. What it means is that these countries have a more developed system of laws criminalizing trafficking, all else being equal.

East European countries do not do better than others in terms of prevention and protection, however (Models 3 though 6). Relatively weak civil societies may keep them from the forefront of prevention (although, importantly, they do no worse here than other countries). The findings may also suggest that pressures from the EU were more about border control than victim protection. In addition, since these countries are more likely to be a source rather than a destination for trafficked individuals, state officials may view this aspect of implementation as less relevant to them than prosecution.

Figure 4 presents the mean value of prosecution index in Eastern European countries and all other countries combined. The rise in the prosecution index in Eastern Europe is also consistent with our statistical analysis presented in Table 2, which shows strong positive effects of the Eastern Europe dummy variable on the prosecution dimension. Compared to all other countries, the sharp increase in the prosecution index in Eastern Europe is notable. Intuitively, this trend suggests the prevalence of human trafficking after the collapse of Soviet Union in 1990s and the significant efforts of these countries to comply with cultural expectations of other countries. For example, countries such as Albania, Bosnia-Herzegovina, Bulgaria, Croatia, Moldova, Romania, and Serbia are the source, transit, and destination countries of human trafficking, and they take several steps to prosecute human trafficking crimes (US Department of States 2010).

Figure 4.

Mean value of Prosecution Index Scores Over Time for Eastern Europe and Other Countries (Range: 0–4)

In terms of other country characteristics, GDP per capita had a consistent positive effect on countries' records of human trafficking prosecution. This is a statistically significant association in every model except Models 3 (prevention) and 5 (protection), where links to international organizations are included. Wealthier countries are more likely to be destination countries for trafficked persons, and this apparently influences their amenability to the formal implementation of trafficking laws.

GDP per capita was an important control to determine whether our observed relationships extended beyond wealthy Western countries, and including GDP per capita in the models did not mitigate the other statistically significant effects. As an additional test of the stability of the other effects, in models not shown, we included a dummy variable of whether a country belonged to the Organization of Economically Developed Countries (OECD). The other relationships remained much the same when OECD membership was controlled, except that this cancelled out the effect of GDP per capita. This is predictable, since GDP per capita is one basis for membership in the OECD. In addition, including both GDP and OECD in the models made the effect of INGOs non-significant, although it did not change the direction of the association.

Democracy had a small but significant positive effect on prevention (in Models 3 and 4) and protection (in Models 5 and 6). Democratic countries tend to be more closely aligned with the world polity; democracy also tends to be associated with more INGOs and lawyers. Including these other indicators in the models is capturing some of the effect that might otherwise be attributed to democracy. In any event, the relatively small size of the effects highlight that more research is necessary to determine precisely when and whether more democratic countries are better at the formal implementation of globally-promoted criminal law.

Our study complements Cho et al.'s (2011b) analysis of human trafficking policy implementation. Like Cho et al. (2011b), we find evidence that policies spread globally rather than being adopted independently by each country. We find that, beyond contagion across proximate countries, the ratification of CEDAW, women in parliament, and the presence of INGOs were positively associated with prevention and protection. Meanwhile, countries with high levels of lawyers' organizations and Eastern European countries scored higher on average on prosecution. Appendix B shows the consistency of our results when a contiguity-weighted spatial lag, such as that used by Cho et al. (2011b), is included. Results from Cho et al. (2011b) and survey research of law enforcement officials in Brazil (Studnicka 2010) suggest that corruption within states may hinder the implementation of anti-trafficking policies. However, our analysis (in models not shown) revealed no effect for corruption on any of the three dimensions of implementation, once world polity factors were controlled. World polity factors also cancelled out the effects of the level of women's economic rights.

As a final note, we included two other relevant indicators in the model to verify our results. We controlled for population (logged) and remittance inflows (worker's remittances as a percentage of GDP), thinking that these might have a functional impact on countries' approaches to migration.¹¹ Neither of these indicators had statistically significant effects on the dependent variables.

CONCLUSION

Our study explored how international factors and domestic characteristics affect the implementation of three different dimensions of a global criminal law regime: prosecution, prevention, and victim protection. Aided by new cross-national, longitudinal data on human trafficking policies (Cho, Dreher, and Neumayer 2011a), we contribute to prior work on globalization and law by explicitly considering the implementation of these different dimensions of human trafficking policies. Our study also creates more possibilities to theorize the space between diffusion and homogeneity (see Grattet, Jenness, and Curry 1998).

We begin this section with some cautionary notes. While the associations in our ordered probit models suggest that certain factors influence policy implementation, they do not prove causality. In addition, the trafficking indexes must be interpreted cautiously. Numeric measures give weight to particular reforms while overlooking others (Davis, Kingsbury, and Merry 2012).¹² They cannot fully articulate the lived experiences of trafficked persons within countries. Finally, on their face, the Cho et al. (2011a) scores distinguish more between little and no reform than between minimal and major reform. Even if all countries scored well on all of the indexes, that would not mean the problem of human trafficking was solved. As with any quantitative measures, where detail is lost, breadth is gained. These data invite an assessment of how human trafficking reforms are progressing globally.

Turning to that question, as expected, we find that dimensions of implementation follow unique trajectories. The prosecution of human trafficking, which is mandated under international law, is stronger initially and sees the most improvement over the period studied. The average of all country prosecution scores increased nearly a full point on a five-point scale from 2000 to 2008. State responsibilities concerning prevention and victim protection are more qualified and ambiguous under international law. For the most part, international law asks rather than requires states to take definitive actions in these areas. Particularly from 2002 to 2008, there was little overall improvement on countries' average score in these areas (although particular countries changed over time). These general trends suggest that "hard" law matters more than "soft" law at the global level, at least in the early years of an international legal regime. Countries meet formal requirements but go little further.

Using ordered probit analyses, we begin to unpack the possible mechanisms that explain these general trends. Importantly, our analyses suggest a more complicated story than simply states' fear of repercussions for violating the mandatory terms of international law. Neither the ratification of the Trafficking Protocol nor U.S. human-rights-related aid was associated with relative compliance with the international human trafficking regime. Countries' that have ratified the Protocol are monitored; they are also linked to aid through the TVPA. If concrete incentives were the primary drivers of state action in this realm, we would expect these factors to matter, particularly for prosecution.

Our finding that prosecution scores improve over time regardless of whether countries are parties to the Trafficking Protocol or dependent on U.S. aid may mean that it is global consensus (at least among states) rather than specific threats or punishments that encourage states to act. This is consistent with prior literature that points to the importance of "institutional effects" at the global level (Schofer and Hironaka 2005; Benavot and Resnick 2006). Scholars have found that global consensus can drive outcomes as much as specific links to international law or the presence of particular national laws, for example, in the cases of environmental reforms and primary school enrollments. Our findings are consistent with the world polity notion that global policy adoption is not simply a matter of coercion or a quid pro quo of reform for resources (see also Boyle and Preves 2000; Kim and Boyle 2011).

Our multivariate analysis also suggests that world polity factors, women's advocacy, and domestic political interest groups matter more for the implementation of the softer elements of the international human trafficking regime. While prosecution, officially mandated by the Trafficking Protocol, was relatively impervious to global ties and domestic interest groups (with the exception of human rights lawyers' organizations), both prevention and prosecution were associated with these factors.

In general, indicators representing concepts from world polity theory (that is, the ratification of CEDAW and the presence of INGOs) were associated with prevention and protection but not with prosecution. Likewise, it is in the two areas less closely associated with criminalization that women's advocacy appears to matter most. We find that the ratification of CEDAW and the level of women in parliament are positively associated with prevention and protection but are not related to prosecution.

We offer several possible explanations for these findings. First, our results may indicate that informal pressures from global and local activists matter more in areas where international law is more qualified and ambiguous (or, put the other way, that that informal pressure matters less when law is certain and direct). This is consistent with the idea that mobilization for change has the greatest impact in areas that are less fully institutionalized.

The results are also consistent with the possibility that activists are focusing their resources on the areas where they see the greatest need. They may perceive that prosecution will move forward regardless of their intervention and hence give it less attention. Another possibility is that women's advocates and states actually have different priorities—that women's advocates are eschewing prosecution because they disagree with that emphasis. The findings may reflect the specific concern that women's social and economic circumstances are a major contributor to trafficking.

The less-formal forms of implementation may be seen as more consistent with world civil society principles. It makes sense that prosecutions and the harshness of punishment could be perceived as running counter to the altruistic spirit of global actors in general (Boli and G. M. Thomas 1999). INGO actors, members of the CEDAW Committee, and women in parliament may look beyond these formal demonstrations of commitment to other factors.

Lawyers' organizations operated somewhat uniquely. Our results reveal that when the number of lawyers’ international human rights-oriented organizations is high within a country, both the criminalization (prosecution) and the prevention of trafficking tend to be stronger. The presence of such organizations had no effect on victim protection, however. Because of their professional training, lawyers may privilege the use of formal law even if this means adopting a more restrictive view of who constitutes a victim (see (Munger 2015). In addition, under certain conditions, lawyers are important for strengthening civil society (Halliday and Karpik 2012). Hence, it makes sense that LPOs are positively associated with prevention.

More democratic countries are more likely to implement prevention and victim protection than less democratic countries. This is consistent with the democratic focus on individual rights and civil-society political engagement. There is no difference in levels of prosecution between more and less democratic countries. In contrast, Eastern European countries have markedly more developed systems of prosecution than other countries. In terms of prevention and victim protection, however, these countries are statistically indistinguishable from others. This finding does not necessarily imply that Eastern European countries are bringing many more trafficking cases than other countries. Rather, it suggests that these countries have a more developed system of laws criminalizing trafficking. The findings suggest that pressures on these countries from the EU were more about border control than victim protection.

In broad terms, our results suggest that global ties and domestic interest groups matter more in areas where international law is less defined. While prosecution, officially mandated by the Trafficking Protocol, was relatively impervious to global ties and domestic interest groups, both trafficking prevention and victim protection were associated with these factors. Our findings call for an elaboration of theories of globalization and law that treats the implementation of law as a multifaceted process.

APPENDIX A

Albania, Algeria, Argentina, Armenia, Australia, Austria, Azerbaijan, Bangladesh, Belarus, Belgium, Benin, Bolivia, Brazil, Bulgaria, Burkina Faso, Cambodia, Cameroon, Canada, China, Colombia, Democratic Republic of Congo, Costa Rica, Croatia, Czech Republic, Denmark, Dominican Republic, Ecuador, Egypt, Estonia, Ethiopia, Finland, France, Gambia, Georgia, Germany, Ghana, Greece, Guatemala, Haiti, Hungary, India, Indonesia, Iran, Ireland, Israel, Italy, Japan, Jordan, Kazakhstan, Kenya, Republic of Korea, Kyrgyz Republic, Latvia, Lebanon, Lesotho, Lithuania, Macedonia, Malawi, Malaysia, Mali, Mexico, Moldova, Mongolia, Morocco, Mozambique, Nepal, Netherlands, New Zealand, Nicaragua, Nigeria, Norway, Pakistan, Papua New Guinea, Peru, Philippines, Poland, Portugal, Romania, Russia, Rwanda, Senegal, Serbia, Sierra Leone, Singapore, Slovenia, South Africa, Spain, Sri Lanka, Sudan, Sweden, Switzerland, Tanzania, Thailand, Togo, Tunisia, Turkey, Uganda, Ukraine, United Kingdom, United States, Uruguay, Uzbekistan, Venezuela, Vietnam, Zambia

APPENDIX B

Table B1.

Ordered Probit Analysis on Anti-Trafficking Policies with spatial lags, 2000–2008.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
	Prosecution			Prevention			Protection
Trafficking Protocol		−0.103 (0.112)	−0.075 (0.115)		−0.036 (0.103)	−0.009 (0.108)		−0.020 (0.099)	−0.017 (0.097)
INGO (logged)		0.096 (0.129)			0.168 (0.113)			0.173 (0.115)
CEDAW		0.365 (0.427)	0.313 (0.418)		0.988*** (0.265)	0.963*** (0.275)		0.659+ (0.337)	0.681* (0.315)
Percent Women in National Politics	0.000 (0.006)	−0.001 (0.006)	−0.000 (0.006)	0.016** (0.006)	0.015* (0.006)	0.016** (0.006)	0.014* (0.006)	0.013* (0.006)	0.014* (0.006)
Legal Professional Orgs			0.017** (0.006)			0.010* (0.004)			0.003 (0.002)
Eastern Europe	0.452** (0.140)	0.486** (0.166)	0.475*** (0.141)	0.130 (0.107)	0.178 (0.124)	0.138 (0.113)	0.050 (0.114)	0.096 (0.141)	0.042 (0.123)
Spatially Lagged Variable (contiguity- weighted)	0.146+ (0.080)	0.185* (0.081)	0.190* (0.082)	−0.141+ (0.073)	−0.111 (0.074)	−0.123 (0.075)	0.028 (0.110)	0.040 (0.111)	0.027 (0.111)
Level of Democracy	0.016 (0.012)	0.014 (0.012)	0.014 (0.013)	0.032** (0.010)	0.022* (0.010)	0.025* (0.010)	0.040*** (0.009)	0.032*** (0.010)	0.035*** (0.009)
GDP per capita	0.209*** (0.056)	0.173* (0.084)	0.197*** (0.059)	0.179** (0.057)	0.124 (0.077)	0.185** (0.057)	0.146** (0.053)	0.076 (0.083)	0.151** (0.058)
Trade Openness	−0.001 (0.001)	−0.001 (0.001)	−0.000 (0.001)	−0.001 (0.001)	−0.001 (0.001)	−0.001 (0.001)	−0.001 (0.001)	−0.001 (0.001)	−0.001 (0.001)
Lagged DV (t−1)	1.243*** (0.074)	1.234*** (0.072)	1.231*** (0.073)	0.939*** (0.089)	0.898*** (0.087)	0.888*** (0.091)	0.994*** (0.086)	0.976*** (0.087)	0.981*** (0.086)
N of Observations	541	541	541	545	545	545	547	547	547
N of Countries	105	105	105	105	105	105	105	105	105
Log Likelihood	−515.506	−511.127	−508.500	−561.685	−559.693	−559.411	−412.506	−412.394	−409.745

Standard errors in parentheses

⁺p< 0.10,

^*p< 0.05,

^**p< 0.01,

^***p< 0.001