Modern Day Slavery: What Drives Human Trafficking in Europe? June 2011 Preliminary Draft (please do not quote) Abstract As Trafficking in Persons gets increasing attention on the international agenda, this paper examines the determinants of human-trafficking victim inflows into 13 European countries based on official data. We develop a gravity model and our results suggest that human-trafficking occurs within well established migrants and refugees corridors, and victims are preferably exploited in host countries with weak institutions and prohibiting prostitution activities. Additionally, liberalization of border controls exacerbates trafficking flows, while no significant effect from asylum recognition rates was found. Our results also indicate that women in equalitarian or repressive societies towards women's rights are less likely to be trafficked.
1. Introduction Declining communication and transportation costs have permitted the intensification of migratory flows during the last decades, triggered by the willingness to locate in a more prosperous and politically stable region. These together with strict border controls and narrow working opportunities for foreigners in host countries, have nurtured businesses assisting illegal migration. Such businesses have not been absent from the involvement of transnational crime organizations, originating the Trafficking in Persons (TIP) phenomenon. This illicit activity, often acknowledged as modern day slavery, bases its source of profits on the exploitation and detriment of human rights by using people as commodities. 1 According to the U.S. Department of State, between 600.000 and 800.000 people fall into trafficking mafias each year, 2 and the International Labor Organization (ILO) estimates that at least 2.4 million adults and children are in forced labor and sexual servitude as a result of trafficking in persons. 3 Moreover, revenues from exploitation are calculated to be not less than $30 billion annually, making human traffic the most profitable illicit activity after drugs and weapons smuggling (ILO, 2005). Despite the magnitude of these figures, quantitative studies on the area are almost nonexistent, mainly because acquiring data is difficult and sensitive. The few empirical studies available are either based on incomplete or dubious data sets or on surveys implemented for specific case studies, limiting their comparability and causing difficulties to come with a consensus on what is driving human trafficking. On the other hand, we are aware that several national institutions 4 in the European Union have systematically collected valuable information on TIP in recent years, which until this point has been largely unexplored. 5 We believe the data reported by these institutions is the most comprehensive and reliable available on this subject and therefore more consistent and precise results can be drawn from it. 1 Modern day slavery is widely used to name the phenomenon of human trafficking and its characteristics today in delimitation to historical slave trade. See for instance Danailova-Trainor and Laczko ( 2010). 2 See U.S. Department of State, Trafficking in Persons Report, 2006. 3 See International Labor Office, A Global Alliance Against Forced Labor, 2005. 4 Complete list of institutions listed on Table 1. 5 This study focuses only on Europe due to data availability and quality. Other important destination countries for victims of TIP outside Europe did not publish comprehensive figures on size and origin of victim flows by the time of this research. 1
With the information provided by these official institutions, we built up a gravity model for human trafficking victim flows between 120 source countries worldwide and 13 host countries in the European Union, with a series average length of 5 years per reporting country. Host countries are in this case the same as those reporting the victim flows, as human trafficking victims can be only identified in the countries where they have been exploited, which are in fact our host countries. In the right hand side of the equation, we added migratory, political and criminality factors in both source and host countries, as well as aspects affecting directly the trafficking business. Among these aspects is legal status of prostitution in the host country, border openness and asylum seeking acceptance rates in a host country towards citizens from a specific source country and migratory and refugee flows within country host-source country pairs. This paper improves quantitative studies on the determinants of human trafficking because, first, it employs official victim flows, which we believe are more precise to describe their configuration and intensity than other estimations or proxies used in previous studies. Second, given the inclusion of multiple host countries, our model allows the analysis of characteristics of countries where victims are exploited, which we found are relevant to shape the direction of TIP flows. And third, as our data set contains information for several years, our study captures the evolution of policies with tremendous impact, such as the expansion of the European Union. Our findings indicate that adverse economic and political conditions and intense criminal activities in source countries fuel international trafficking flows. Additionally, from the host country side, prohibiting prostitution activities, liberalizing border controls, and low asylum seeker acceptance ratios exacerbate the problem. The results also indicate that women from highly repressed and from liberal societies regarding women s rights are less likely to be trafficked. The paper is structured as follows: Section 2 presents an overview of relevant studies on the topic and highlights the position of our analysis in the literature. Section 3 describes the phenomenon of human trafficking and its delimitation with illegal immigration and smuggling. We state our different hypothesis in section 4, illustrating how characteristics of host and source countries can facilitate the contact between potential migrants and smugglers of persons, and also how these can give incentives to criminal organizations to engage in this kind of illicit activity. In Section 5, the empirical identification strategy is 2
described, as well as the data and the model used. In this section we also exhibit results obtained, revealing which factors are crucial in the determination of TIP in Europe. Finally, in section 6 we summarize and draw conclusions. 2. Literature review As mentioned, quantitative research on human trafficking is scarce. However recent attempts have been made to address this subject empirically, due to the growing importance of human trafficking on the international agenda. One part of literature analyzes the impact and effectiveness of the implementation of policies combating human trafficking (Akee et al., 2010b; Avdeyeva, 2010; Cho and Vadlamannati, 2011; Simmons and Lloyd, 2010). In line with that strand of the literature Cho et al. (2011) develop an index to measure countries overall policies on the three main policy dimensions prosecution, protection and prevention. A second strand focuses on the factors driving TIP. We take a closer look at these studies and highlight our contribution. One of the first studies is conducted by Akee et al. (2007) who examine determinants of women and children trafficking flows worldwide by employing a gravity model. They predict a self constructed binary variable which indicates if at least one individual has been trafficked within a source-host country pair. Besides inequality and political instability in source countries, this study finds that TIP is also exacerbated by banning prostitution in source countries and granting legal status to victims in host countries. These last two should increase returns on prostitution and incentives to illegally migrate to that host country respectively, triggering flows. In a separate more recent work, the authors address the drivers of all kinds of human trafficking by constructing an extended gravity model. Here they focus especially on the impact of prostitution laws and legal support systems for victims in both host and source countries. By employing a simple probit model as well as an instrumental variable approach, they find that granting legal status to the victims in the host country positively and significantly influences the probability of trafficking. Only some specifications indicate a positive impact of stricter prostitution laws in the host and source countries on trafficking 3
fows (Akee et al. 2010a). 6 Focusing in the influence of ethnic fragmentation, conflict and internally displaced persons and international trafficking, Akee et al. (2010b) find that conflict-prone countries with large numbers of internally displaced people and refugee outflows tend to also be source countries of trafficked victims. Another victim flow gravity model is constructed in Jac (2010). However this time, the analysis is restricted to victim inflows into the US only, therefore concentrating mainly on source countries conditions that promote human trafficking. The study utilizes the number of T-visas granted to relatives of victims by country of citizenship as proxy for inflows. 7 She finds that geographical proximity and migrant networks explain TIP flows into the US, while economic and social disparity between host and source countries both have no significant influence. Taking into account some of the factors analyzed in the previous studies the contribution of our work is threefold: First, it employs official victim flows, which are more precise to describe their configuration and intensity than other estimations or proxies used in previous studies. Second, given the inclusion of multiple host countries, our model allows the analysis of characteristics of countries where victims are exploited, which we found relevant to shape the direction of TIP flows. And third, as our data set contains information for several years, our study captures the evolution of policies with tremendous impact, such as the expansion of the European Union. In this way, the assessment of host and source country characteristics within a human trafficking frame is improved. 3. What is human trafficking? A first step to conduct quantitative studies is to have a clear concept of the phenomenon. The most basic definition of human trafficking refers to the smuggling of an individual for the purpose of exploitation. However, this basic definition leaves significant space for interpretation and also overlaps with the concept of human smuggling. For this reason, we address a more specific and formal definition before starting with the identification of determinants. 6 In the instrumental variable approach they address the concern of endogeneity of victim protection programs and country specific unobserved heterogeneity that may be present in the probit specification. They instrument by using legal origin and settlers mortality in order to address these issues (Akee et al. 2010b). 7 T-Visas are visas granted to human trafficking victims in order to get legal status in the US. The regulation was issued by the Department of Justice in 2002 and is an initiative to get victims report to officials. The author uses the number of T-visas granted to relatives of victims for confidentiality reasons (Jac 2010, 15). 4
3.1. Definition and distribution The most precise definition of TIP is given by the Protocol to Prevent, Suppress and Punish Trafficking in Persons also called the Palermo Protocol, issued in December 2000 and adopted by the UN. 8 This protocol is the most serious attempt to harmonize the definition on human trafficking internationally. In Article 3 it describes TIP as the recruitment, transportation, transfer, harboring 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. Specifically, exploitation shall include, at a minimum, the exploitation of the prostitution of others or other forms of sexual exploitation, forced labour or services, slavery or practices similar to slavery, servitude or the removal of organs. (UN, 2000a). This definition highlights three core components of trafficking, namely, the use of deception or coercion, the movement or harboring of an individual, and its placement in situations of exploitation. These 3 elements must occur consecutively to recognize a TIP case. According to the Global Report on Trafficking in Persons (UNODC, 2009), trafficking in women and girls for the purpose of sexual exploitation is the most common case of human trafficking, accounting for 79% of all cases identified, followed by trafficking for the purpose of labor exploitation, with 18% of all cases identified. Moreover 66% of all human trafficking victims are women, 13% girls, 12% men and 9% boys (ibid: 11). This indicates that more than a fifth of all victims are children. These figures could, however, misrepresent the reality of the phenomenon, and the distribution of the victims might reflect, in fact, the attention of the international community to specific segments of TIP (ibid: 7). 3.2. Distinction between illegal immigration, smuggling and human trafficking Human trafficking is considered a variation of human smuggling. In Article 3 of the Protocol against the Smuggling of Migrants by Land, Sea and Air: Smuggling of migrants means the procurement, in order to obtain, directly or indirectly, a financial or other 9 8 The Protocol to Prevent, Suppress and Punish Trafficking in Persons is supplementing the United Nations Convention against Transnational Organized Crime together with the Protocol against the Smuggling of Migrants (UN, 2000b). Both were issued in December 2000. In addition, the UN inaugurated a Global Initiative to Fight Human Trafficking in 2007 (UN.GIFT). 9 For this same reason, the large majority of studies on TIP focus on trafficking of women and girls for the purpose of sexual exploitation, as there are more information available for this segment. 5
material benefit, of the illegal entry of a person into a State Party of which the person is not a national or a permanent resident. (UN, 2000b). As seen, both definitions of human trafficking and human smuggling overlap in some degree. In order to explicitly identify and separate the phenomenon human trafficking from illegal immigration and human smuggling, a clear delineation is given by a nested concept, as has been suggested by Väyrynen (2005). According to him illegal migration, smuggling and trafficking are nested concepts with a shared illegal character of the entry to a country. In other words, illegal migration or crossing borders violating immigration laws and voluntarily staying in the country of destination is an essential component of both human smuggling and trafficking. Nevertheless, the economic deliberations, as well as the steps undertaken by the agents involved, are distinct. The basic matter of smuggling is the assistance on border passing. The smuggler receives a financial remuneration before the journey and acts as a support to illegally cross over international borders. Most importantly, the smuggled individual is free in the destination country. In human trafficking, however, the intermediate person does not only earn from transporting the individual but also from exploiting or selling him in the destination country. This is possible through physical coercion, fraud or deception at any point of the process, and eased when a financial remuneration is still expected by the smuggler after the individual has arrived at the destination country. Now the two concepts of smuggling and trafficking in persons seem to be straightforward. However, it is important to highlight that in many cases people get trafficked without being assisted to cross an international border illegally. People can enter the country of destination legally with an official document after they have been promised a job as waitress, model or construction worker, for example, and only realize their victimization after the arrival. Upon their entrance they usually have to handover their legal documents and are forced to work in the prostitution business and/or face slave-like conditions. In other cases, the individual can enter the country with a valid short term visit document and accepts a job offers that requires him to stay illegally after the document expires. In such situations, the individual has to pay exorbitant amounts of money to the employer for the voyage and/or to get the documents back, which means he is consequently exploited (Aronowitz, 2001: 166). In these instances of human trafficking, when individuals cross an international border legally, it is even more difficult to detect it. 6
Thus, there are three possible scenarios imaginable in which people might get trapped in the hands of traffickers: In the first, an individual wants to migrate and contacts a smuggler to enter the destination country illegally and eventually ends up trafficked. In this case, having a debt with the smuggler increases sharply the chances of being trafficked. In the second scenario, the individual is deceived by a job offer abroad, which seemed authorized, and his documents are expropriated after his arrival, ending up trafficked. During the hiring process, traffickers collect information on his relatives and friends to manipulate their victim and therefore facilitate his exploitation. And in the last one, the individual is kidnapped in his home country and transported against his will to a different place where he is exploited. 10 These possible modus operandi of traffickers play an important role in identifying factors that determine trafficking in persons. In the next section we closely examine aspects that might influence human trafficking according to anecdotal evidence. 4. Determinants of human traffic and hypothesis We identify factors that influence TIP, taking into account migration pressure aspects as well as incentives for trafficking mafias to surge and operate. On the one hand, human trafficking falls under a broad concept of migration, as the willingness to migrate is an essential condition for trafficking to occur. On the other hand, individuals can only be trafficked if criminal organizations are present at any point of the migratory process. We examine push and pull factors or conditions in source and host countries shaping and intensifying migratory flows, as human trafficking organizations make use of existing migration routes to minimize operation costs. Push factors increasing the pressure for migration include low living standards, lack of economic opportunities, political repression and demographic growth. The higher the migration pressure, the more likely an individual will get in contact with a trafficking mafia by accepting dubious means to settle abroad. Pull factors, measuring prospects of better living conditions in host countries, comprise economic opportunities, political freedom and availability of land. Victims are 10 An additional way of being victimized can be imagined where parents send their children away in the hope that they would have better living or economic conditions. This takes mainly place within countries and is not captured in our data, as we cover international victim flows. 7
more likely to end up exploited in countries preferred to migrate to, because for trafficking mafias it is costly to avoid migratory patterns. 11 In addition, we examine factors facilitating trafficking operations in source and host countries and which also affect mafias profits. Naturally, these criminal organizations prefer to recruit victims where it is less costly, transport them through a less risky route, and exploit them where they can obtain higher revenues. Transnational mafias surge in countries with low governance levels and weak judicial systems, and in the case of human trafficking, prefer to transport victims within short distances. From the profits side, exploitation yields in specific niche markets, such as the sex industry, are the most relevant. These migratory and business factors are classified into four sets of variables: migration factors, political environment, criminality prevalence and human traffic specific factors. The subsequent choice of variables is based on the analysis of these factors of sending and receiving countries. Out of this analysis we derive hypothesis which are then going to be tested empirically. 4.1. Migration factors Expectation of higher incomes, better living conditions and employment opportunities abroad encourage the decision to migrate. When the income gap is high enough, people are inclined to take the risk to migrate even if they have a stable source of income in their home country (Harris and Todaro, 1970). Moreover, generous welfare systems in high income countries can even attract individuals without a certain job offer in the recipient country (Rotte and Volger, 1998; Giorgi and Pellizzari 2009). From the other side, in high income economies the demand for foreign cheap labor in insecure and informal service sectors such as prostitution, domestic and factory work is substantial; as such openings are barely filled by nationals. 12 Given this market fragmentation, together with restrictive legalization opportunities 13, migrants are left with no other chances than subsisting in the shadow economy (Friebel and Guriev, 2004; Akee et al, 2007). 11 For a detailed analysis of push and pull factors influencing the migration decision of households refer to Stark (1991) or Castles and Miller (2003). 12 These insecure and informal positions are also called 3-D jobs: dirty, degrading and dangerous (ILO, 2003: 4). 13 Since 1992 European countries limited the accessibility to their territory especially for asylum seekers and refugees beyond existing restrictions impeding economic and non-economic migration (Zaiceva and Zimmermann, 2008). The role of this population of migrants will be further explored in section 4.4. 8
Large wage differentials between host and source countries and the employment informality of migrants offer significant profit opportunities to criminal organizations by trafficking people. Exploitation in informal services, language barriers and information disadvantages as well as the fear of alien s deportation circumvent in many cases that victims look for legal support, making it easier for traffickers to keep their victims under strict control. Such profits of exploitation calculated in industrialized countries alone amount for around US$ 15.5 billion per year, which is half the profits gained worldwide (Belser, 2005: 14). On the cost side of the trafficker s calculation, the distance of the transportation process extensively determines expenses. The shorter the journey of transportation the lower the costs and the risks making close by but economically different countries favorable host-source country matches of trafficking victim flows. Moreover, the travelling costs of migrants reduce by shorter distances increasing the number of migrants and subsequently the vulnerable population in the host countries that traffickers draw on for exploitation. Clearly, as has been extensively explored in the migration literature supply and demand for migrants is over and above determined by the size of the population, the larger the population in a sending country the larger the number of people originating from there and vice versa for the receiving country (Greenwood, 1975; Borjas, 1987, 1989; Karemera et al, 2000). In the trafficking scenario where traffickers are involved in border crossing assistance, facing a larger source country increases the recruiting possibilities and a larger number of people who want to emigrate. Due to this evidence we test the following hypothesis: Hypothesis 1a: Income differentials and disparity in living conditions, larger population size as well as low distance increase the number of victims between countries. 4.2. Political environment The political environment obviously influences the operation of traffickers and thus TIP flows in both host and source countries. Mafias locate their operations where the probability of detection and prosecution is low. In countries where the enforcement of legislation is weak and illicit activity can take place easily trafficking mafias are able to locate without high risks of exposure. In the trafficking process this is true at both ends of trafficking routs. The legal enforceability is guaranteed by the effectiveness of the prevailing 9
institutions and the judicial system to prosecute traffickers. This induces not only a risk factor, hence increases costs for criminal activity, but also an incentive for price rises and subsequently higher profits. A qualitative high institutional environment, additionally, is the basis for the provision and enforcement of legal rights of the population. Repression and no political freedom existing in many countries of origin of migrants, increases further the migration pressure and plays into the hands of traffickers. As the great amount of victims is women (see section 3.1), an important driver of the choice to (illegally) migrate is the situation of these women in their home country. Given the possibility to migrate, i.e. by a promising job offer, and overcome the existing repression against their own persons, incentives women to contact intermediaries to help them leave the country. In order to capture this situation, we control for the rights of women guaranteed in the country of origin. In countries where women are very repressed we expect them to have no possibility to contact smugglers while in more liberal countries this should increase. Considering this substantiation the hypothesis investigated is twofold: Hypothesis 2a: Mafias operations are facilitated in countries with unstable legal systems, increasing the number of victims. Hypothesis 2b: rights for women. Women are less likely to be trafficked from countries with equal 4.3. Criminality prevalence TIP is a large scale international business operating at all steps along the way supported by intermediaries (traffickers) that smooth movement between origin and destination country. The three steps in the trafficking process are: the mobilization, in which people are recruited in their home country; implementation, where people are transported; and the insertion and integration of the individuals concerned in the destination country (Salt and Stein, 1997). In order to arrange this kind of international business organized criminal activity is supposed to play a major role (Stoecker, 2000; Salt, 2000). Accordingly, organized crime active in human trafficking can be characterized as a number of small organizations that are highly flexible and that can thus be modified as required. (Brucker and Parent, 2000: 17). Therefore, a close connection to other kinds of illicit trafficking such as drug or arms trafficking is likely to exist as the structure of enterprises (legal or illegal) 10
depends on the interrelation within and between organizations, thus diversification is one of the major goals. This applies especially for illegal organizations due to the risk of legal exposure and the perspective of prosecution. Therefore, a connection of drug and arms trade and human trafficking seems plausible, thinking of the well established trafficking routs and contacts along the way already existent (Schloenhardt, 2001). Nevertheless, due to the fact that a definition of the concept of organized crime is difficult to obtain, especially in a multi-country context, and that the measurement of illicit activity in general is based on estimations, the assessment of the scope of the problem is difficult to accurately judge (Bruckert and Parent, 2000: 13). However, looking at the involvement of organized crime and how much this aspect drives human trafficking in Europe is essential in assessing the phenomenon. The prevalence of criminal activity in the countries captures lower operation and information costs for the criminal organization active in the shadow migration market. By controlling for the level of criminality in the countries the magnitude of the problem will be addressed and its influence on the number of victims can be tested. Hypothesis 3: Trafficking mafias surge easier and locate their operations in countries with high criminality levels. 4.4. Human trafficking specific factors The following set of factors influence particularly the modus operandi and profits of traffickers. We are considering: prostitution law, migration networks and refugee flows, asylum policy and border controls. As has been outline in the last section the trade of humans is an economic activity controlled mainly by organized crime with agents acting as businessmen trying to maximize profits. Furthermore, a great amount of victims ends up to be exploited in the sex industry, but only where market conditions are profitable to sell people. These two aspects together indicate that the characteristics of the national sex market for commercial sex play a major role in the decision on the destination country. As Jacobson and Kotsadam (2010) put it, the official framework of the prostitution industry is crucial in the determination of profits. There are two lines of argument on how the legal framework on commercial sexual services might influence the number of human trafficking victims: On the one hand, authors find that victim numbers are higher in countries where prostitution is legalized (Jakobson and 11
Kotsadam, 2010). Consequently, sanctioning prostitution generally will decrease the problem of trafficking within the country. On the other hand, exactly the opposite can be argued, or that trafficking can only be reduced if prostitution is legal due to lower prices for sexual services, which reduces incentives to participate in the trafficking business significantly (Aghatise, 2004; Akee et al. 2010a). Second, as we highlighted before, migration is a key component of human trafficking, where criminal organizations analyze costs and benefits of this natural movement of people within their trafficking business. Abducting a person and smuggling him in to a country favored by the trafficker 14, is more costly than, for example, localizing trafficking activities only in both ends of a specific migratory corridor. 15 In this way, operation costs are reduced, as trafficker s risks associated with border crossing are minimized. Additionally, traffickers can make use of existing migration networks through the misuse of trust relationships between friends and relatives to the migrant s disadvantage (Mahmoud and Trebesch, 2009). As Gozdziak (2009: 165) points out, these smaller operations based on kinship or friendship ties may, of course, be part of larger criminal networks. For these reasons, traffickers benefit from locating their activities within migratory flows, and therefore the number of TIP victims should increase the more intense such flows are. Similarly, refugee flows are also attractive for human trafficking businesses. This population, besides presenting the same characteristics as migrants, is also willing to give up certain liberties and even accept being exploited in order to reach a certain destination. Therefore, traffickers will be highly interested in the origin and destination of refugees. Third, policies carried in host countries regarding asylum seekers should affect the vulnerability of the refugee population. The more restrictive a country is in granting asylum rights, the higher the probability that this population ends up in illegality and subject to exploitation, as refugees try to remain in the destination country at any means. In addition, Friebel and Guriev (2004) can theoretical show that the stricter immigration laws of countries the higher the inducement of traffickers to target these countries, due to increasing profit expectations. This suggests that victim numbers should be larger in countries with highly restrictive asylum seekers policies. 14 This is the third trafficking scenario described in section 3.2. 15 This is the second trafficking scenario described in section 3.2. 12
Finally, we expect border controls to influence traffickers decisions. However, as in the case of prostitution law, the direction of the effect is ambiguous. On the one hand, when an international frontier is closed, a trafficking opportunity is created because undesired immigrants will look for assistance to cross the border illegally. Mahmoud and Trebesch (2009) support this idea and argue that closing borders force people into irregular migration, leading them to contact trafficking mafias to reduce crossing risks. But on the other hand, when an international border is open, traffickers, basing their business in deceiving migrants with false job opportunities in host countries, will face fewer restrictions to enter in contact with potential victims. Visa requirements for third nationals to enter host countries captures how open borders are for them, and their impact on the number of victims will show which of both effects just described is more prevalent for the European case. As has been shown the additional factors drive human trafficking in several ways. The following hypotheses are tested in order to address these factors and estimate their influence on the victim flows within Europe: Hypothesis 4a: Legislation on prostitution impacts traffickers profits via supply and demand for sexual services. A priori, the direction of the effect on the number of victims is ambiguous. Hypothesis 4b: Border controls affect trafficking flows in an indecisive way: closing borders increase illegal migration, but opening borders decreases trafficking costs Hypothesis 4c: Migratory and refugee flows are an essential precondition for trafficking to occur. Hypothesis 4d: Restrictive asylum recognition rates promote illegal migration, and therefore trafficking. 5. Empirical analysis In this section we describe the data sources, present the identification strategy, test the hypothesis introduced in the previous section and show our results. 5.1. Data Our goal was to obtain official numbers on inflows of human trafficking victims into Europe. For this purpose, we contacted 37 national institutions across countries in the region 13
collecting any kind of statistical information on TIP. 16 Among these were anti-trafficking centers, national police offices, ministries of interior, immigration offices, service centers, research centers and universities, which we asked for information on the size and origin of victim inflows into the reporting country per year. From the 37 institutions contacted, 22 provided us with the information requested, 10 with incomplete or unsuitable data, and 5 did not have any kind of statistics regarding the size or origin of the victims at the moment of contact. 17 It is important to point out that there is not any kind of institution collecting accurate human trafficking data in various European countries, such as in France and Italy, which is the reason why no entity was contacted in these countries. 18 A list of the institutions contacted, as well as a short description of the data provided can be found in table 1. Finally we obtained the size and origin of TIP inflows of 13 countries in Europe. Given that the information provided by the entities consists in the number of victims identified, we carefully looked for data reporting characteristics which we consider might affect the numbers collected. It is of highly importance to keep in mind these characteristics at the time of analyzing and comparing victim numbers across institutions or countries, so it is clear that flow intensities perceived also depend on the quality of the data. We distinguished four dimensions capturing different reporting characteristics among entities. The first dimension is the type of institution collecting the data, which could either be a service centre or a public institution (national police offices, ministries of interior, immigration offices, etc.). We believe that service centers tend to report fewer victims than public institutions, as they include numbers of victims they shelter, which are in fact constrained by their assigned budgets and accommodation available. The second dimension consists in the types of exploitation the source is reporting, some include merely victims of sexual exploitation, while others account for additional types (labor, organ removal, etc.). Naturally in the second case the numbers reported are more likely to be higher. The third 16 Several studies (Laczko, 2002; Kelly, 2002 and Tyldum and Brunovskis, 2004, 2005) mention limitations regarding data quality due to the illicit character of the phenomenon. However, we address this problem by observing data quality characteristics for every source, and control for such features in our identification strategy discussed in this subsection. 17 Institutions that did not provide complete data are for instance, the Federal Criminal Police Office of Germany that publishes only the 10 largest source countries of victim inflows; the National Agency against Trafficking in Persons of Romania publishes victim outflows instead. 18 In France, the Central Office for the Suppression of Trafficking in Persons produces statistics on victims of soliciting prostitution and the Central Office for Fight against Illegal Migration on victims of labor and living conditions against human dignity, both statistics do not refer exclusively to TIP. For Italy, statistics on TIP victims collected by the National Anti-Mafia Bureau are not available to the public. 14
dimension reveals whether nationals are included in their numbers. Even though our study takes into account only international flows of victims, we believe that sources reporting victims exploited in their country of citizenship must be more efficient at identifying victims in general. The forth and last dimension distinguishes how the source is counting the victims. Figures could either reflect the number of confirmed victims or the number of possible victims (still to be confirmed). In this case, sources reporting confirmed victims, which is desirable, present smaller numbers, as not all persons suspected of being TIP victims turn out to be actual victims. Best data quality comes from public institutions reporting confirmed victims of all types of exploitation and including nationals in their numbers, meaning that the reporting entity is fulfilling all 4 dimensions. In graph 1 the average number of victims per year in our 13 host countries is depicted. 19 Here the UK and the Netherlands (fulfill 3 of 4 data dimensions) might be presenting higher yearly averages than Greece, Belgium or Austria (fulfill all 4 data dimensions) partly because these first 2 countries count the number of possible victims while the other 3 countries count the number of confirmed victims. Similarly, Czech Republic, Poland, Sweden and Denmark (fulfill 3 of 4 data dimensions) display lower yearly averages than Greece, Belgium or Austria, partly because these 4 countries do either take only sexual exploitation cases into account or exclude nationals in their reports. This clearly shows the need to control for these four data quality dimensions while doing regression analysis, in this way the comparability of different data sources is guaranteed. 5.2. Identification strategy The identification of drivers of human trafficking in Europe requires an assessment of all factors that might influence the flow as outlined in part 4. We employ a gravity model to account for them. 20 Our variable of interest is the number of human trafficking victims per year originating from source country j and exploited in host country i. The gravity model contains 13 host countries and more than 120 source countries, covering the time frame 19 In case of multiple sources for a country, series with best data quality is chosen, evaluated by the number of data quality dimensions fulfilled. 20 The concept draws on the findings in the trade literature identifying determinants of foreign direct investment and has been applied to analyze migration flows. See for instance Anderson and Wincoop (2003) for the usage on drivers of trade flows and Sjaastad (1962), Greenwood (1975), Borjas (1987, 1989) and Karemera et al (2000) in the migration literature. 15
1998-2009. 21 The estimation method employed is a zero inflated negative binomial model given that our objective variable is over dispersed and contains a large proportion of zeros. 22 On the right hand side of the equation we added economic, political, criminality and human trafficking specific factors to test the hypothesis discussed in section 4. Migratory factors (M) include GDP per capita in host and source country, population in host and source country and geographical distance between host and source country. Political variables (P) embrace a law and order index 23 in host and source country and women s political rights index in source country. 24 Criminality factors (C) are homicide rates in both host and source country. Human trafficking specific factors (H) comprise the following 5 variables: First, legal status of prostitution in host country, taking a value of 0 if any kind of prostitution activity is illegal, 1 if only the exchange of sexual services for money is permitted, 2 if additionally pimping or operating brothels is allowed, and 3 if all kinds of prostitution activities are legalized. Second, visa entry requirements in host country i for citizens from source country j, which takes a value of 0 if a visa is required to enter the country for any motive, 1 if no visa is needed to enter the country for short visits, 25 and 2 if no permit is needed to remain in the country permanently or even work. The third variable is the number of citizens from source country j residing permanently in host country i for a given year. Forth, number of refugees originating from source country j and residing in host country i for a given year. And finally, the acceptance rate of political asylum seekers coming from source country j and applying in host country i. 26 We also include four dummies recognizing the fulfillment of the four data quality dimensions (Q) mentioned in the previous section, and add host country, source country and time fixed effects ( respectively). For a list of our data sources please refer to table 2. The complete specified gravity model can be seen in the following equation: 21 For time series length per data source, please refer to Table 1. 22 The proportion of zeros is around 70% for our variable of interest in the 4 specifications in Table 3. 23 Law and Order index by ICRG was chosen since we are interested in how operations of human trafficking mafias are facilitated/obstructed by partial/impartial legal systems. Additionally, this index presents a correlation of just 0.2 with GDP per capita in our data sample. 24 The squared term of this variable was included as well. 25 A short term visit cannot exceed 3 months according to the Council of the European Union. 26 This ratio is equal to the number of asylum applications approved within the year of interest, divided by the total number of decisions taken (approvals, rejections and cases closed) during that same year. 16
5.3. Results The estimation results for different specifications are shown in table 3. The first specification includes data quality dummies and excludes host country fixed effects and the second one does exactly the opposite. The reason for this, is that since our data quality dummies do not vary over time, their effect is being captured by the host country fixed effects too, making both set of dummies highly correlated. In the third and forth specifications we include criminality variables to the first two specifications. We decided to run the models with the criminality variables separately because unfortunately these have lots of missing values. In these cases, the sample size drops to one third of that of the original specification, therefore we lose a large proportion of relevant information. As expected, income levels in host and source countries and geographical distance between them are important to explain human trafficking flows. Income levels do not only reflect migratory patterns, but also designate where traffickers prefer to locate their operations. Higher incomes in host countries are more attractive for traffickers to exploit their victims due to larger profit expectations, while higher incomes in source countries make it more difficult for traffickers to find potential victims. In addition, traffickers prefer to recruit victims the nearest possible to the chosen host country, as the clandestine transporting of people is highly costly. Furthermore, the size of the population in host and source countries does not matter for traffickers, presumably because the human trafficking market is so narrow that in any country, regardless its size, enough potential customers or victims could be found for the business to function. Moving to the political factors, stability seems to matter more in host countries than in sources countries, although in either case the results are not robust. Traffickers choose to exploit their victims in countries where they are less likely to be prosecuted and keep their operations under cover, viable when the judicial systems are inefficient or when traffickers have some degree of compliance with local officials. Women s political rights were found to be highly important to shape victim outflows. However, they display an inverted u relationship rather than a linear one. This implies that women are less likely to be trafficked from countries which either repress or equalize their political rights. In societies depraving women from their rights, they are regularly not allowed to take decisions by themselves, such as crossing international borders, which in fact prevent them from contacting trafficking mafias. 17
Regarding criminality factors, countries with higher prevalence of criminal activities are more likely to be a source of victims, as expected. This result corroborates that human trafficking mafia s benefit from networking with other type of criminal organizations for operation and information costs reduction purposes. Analogously, criminality levels were found to be insignificant in host countries, suggesting that human trafficking mafias operate by themselves when finally exploiting their victims. Associating with other criminal groups is rather difficult and costly in host countries given the presence of good institutions. Results for human trafficking specific factors show that legalization of prostitution reduces victim numbers in host countries. This is in line with the findings of Akee et al. (2007, 2010a), that banning prostitution in the receiving country increases the probability of trafficking. Prostitution prices drop and thus traffickers profits when the market for sexual services is regulated, due to a positive shift in the supply curve motivated from lower operation costs. Additionally, individuals interested in working in the sex industry do not have the need to contact criminal organizations in an environment where prostitution is completely legal, decreasing their probability to end up trafficked. Interestingly, these results also suggest that relaxing border controls (i.e. enlargement of the European Union) facilitates human trafficking. As mentioned in section 4.4, borders do not need to be closed for human trafficking to occur. In the European case, opening the borders reduces abruptly costs for traffickers to bring victims from east to west, given that Eastern Europe has been traditionally the major source of victims. Continuing with the following factors, bilateral flows of migrants and refugees were found to be highly significant to determine TIP victim flows across all specifications. These results are not surprising, as trafficking mafias have larger chances to recruit victims within populations willing to relocate. As stressed in section 3.2 mafias locate their operations at both ends of these flows in form of businesses assisting legal or illegal migration. It reflects also the fact that traffickers take advantage of trust relationships among migrant networks to deceive potential victims. Finally, lower acceptance rates of asylum seekers were found to intensify victim flows, probably via the promotion of illegal migration. However its effect was not robust, as it was significant only in two of the four specifications. 18
6. Conclusion This study is the first attempt to analyze the determinants of human trafficking by employing official records on victim flows. It is a step forward to the existing literature, since the utilization of data collected by national institutions was completely neglected until now. Naturally, such statistics capture only a portion of the whole phenomenon, as for any other criminal activity, however, it is the best data available to describe the direction and intensity of trafficking flows, and it should be preferred than any other type of estimation or proxy. Our results suggest that adverse economic, political and security circumstances are advantageous for mafias to find potential individuals to be trafficked internationally. These not only reflect emigration pressure, but also ideal conditions for traffickers to establish and maintain their recruitment processes in a certain country. Trafficking mafias benefit from legal systems that reduce their likelihood to be prosecuted, and also from networking with other criminal organizations to minimize operation and information costs. Interestingly, we also found that policies carried out in high income countries, logically preferred to exploit international victims, do shape victim flows. In other words, some countries are more attractive or profitable for trafficking mafias than others to exploit their victims. Legalizing prostitution reduces considerably the room for exploitation, as prices for sexual services drop and therefore profit realizations. Additionally, in such scenario, individuals interested in working in the sex industry can avoid entering in contact with criminal organizations, decreasing their probability to end up exploited. We also found that the transportation reduction effect caused by opening international borders is significantly larger than the illegal migration promotion effect caused by closing them. This means that relaxing border controls is facilitating traffickers to bring victims from source to host countries. However this finding should not be generalized to other regions, as the expansion of the European Union towards the east is a very particular phenomenon. And lastly, we found that highly restrictive policies towards granting refugee status to asylum applicants intensify trafficking flows, however, this evidence was not statistically robust. Trafficking in Persons is a criminal activity driven by profits. Revenues originate mostly during the exploitation phase and policies implemented in host countries can help to shrink them. In particular, the regulation of markets where victims are often exploited, such as for sexual services, agriculture or domestic assistance can considerably reduce the 19