It’s a hard knock life the survival strategies of Dutch Mulas in Peruvian prisons

Not much research has been done in the Netherlands on the ways in which Dutch women experience and survive a prison sentence abroad. The research underlying the present article focused on Dutch female prisoners in foreign countries, using an ethnographic research method. The article looks at the prison experiences of Dutch mulas in Peru and their strategies of survival. The ways in which these women react to their detention can be traced back to existing literature on the importation theory and the deprivation theory. The article also examines the various coping strategies that are used to survive a prison sentence abroad. These strategies appear to be related both to the life experiences of the women (importation theory) and the experienced prison environment (deprivation theory).


Introduction
A relatively large amount of research has been conducted on the prison experiences of women [1][2][3][4][5][6][7][8][9][10], but as far as we know, very little field research has been done by researchers on the experiences of women incarcerated in a foreign country. The research on which this article is based deals with Dutch women detained in prisons in Peru. This ever-growing group of detainees is faced with surviving a prison sentence far from home. Although no exact figures are available, the involvement of women in drug smuggling worldwide has reportedly increased over the past several years [11,12]. In 2005, there were thirteen Dutch women in Peruvian prisons; in 2010, the period of our research, there were twenty-five [13,14]. In April 2012, the number had fallen to eleven as a result of sentence reductions of up to 9 years.
Prison life in Peru is very different from life in a Dutch prison and Dutch inmates in Peru are faced with all sorts of problems that are usually non-existent in their own country. For instance, there is often a language barrier, most Dutch prisoners receive no visits from relatives, some prison officials may be corrupt, and standards of hygiene and privacy are often lower than in The Netherlands (Ministerie [15]). In short, the conditions in foreign prisons are likely to be tougher than those in prisons in The Netherlands. The present article aims to examine how Dutch women faced with such circumstances manage to endure their incarceration. This study was part of a larger research project on Dutch women incarcerated abroad.

Survival in theory
With the exception of Janine Janssen [16], who studied South-American mulas 1 in the Bijlmer prison in Amsterdam, ethnographic methods are almost never used in prison research in the Netherlands and, as far as we were able to ascertain, this type of research has never been conducted on Dutch citizens incarcerated abroad. This is all the more remarkable given that ethnographic research can provide a rich insight into prison life. Janssen's research is particularly interesting. Over the course of several months, she spent many hours with nine Latin American women who were detained in a Dutch prison after being caught smuggling cocaine at Schiphol Airport. Using participant observation and in-depth interviews, Janssen was able to show that both the importation and the deprivation theory were applicable to the prison experience of Latinas in the Netherlands. The importation theory holds that the manner in which people cope mentally with detention is determined by their way of life prior to their imprisonment [10,17]. Supposedly, detainees are more often victims of painful life experiences, such as physical and/or sexual abuse, than people without a criminal record [18][19][20][21]. Green et al. [20] for example state that Bthere is an increasing understanding of the fact that many women in prison or jail for committing crimes are crime victims themselves. In particular, they have often been abused sexually and physically in childhood, many have been raped, and exposure to family violence is endemic.^According to Adams [22] people develop ways of dealing with painful experiences and they take these strategies with them into the prison system. Variables such as a criminal record, cultural background and mental health, as well as personal characteristics such as ethnic origin, gender and age, are also imported into the prison and these factors will determine to a certain extent how an inmate deals with his or her sentence [7,[22][23][24]. However, a prisoner's personal characteristics and life history are not enough to explain a person's response to imprisonment because, as argued by Zamble & Porporino [24], human behaviour is rarely determined by personal characteristics alone. It is much more often the result of interactions between an individual and his or her circumstances. In line with this argument, prison conditions are also considered important in determining how an individual responds to his or her imprisonment. The deprivation theory holds that a person's response to a prison term will be determined by the challenges imposed by the prison system and by circumstances over which the prisoner has no control [10,23]. In Gresham Sykes' groundbreaking study 'The Society of Captives ' (1958), these challenges are referred to as the pains of imprisonment. Restrictions to the freedom of movement lead to social deprivation: for instance, the prisoner is no longer able to maintain social ties with family and friends. He is also limited in his freedom to receive goods and services and is deprived of heterosexual relationships. Sykes also mentions the loss of autonomy and the loss of security as consequences of a prison sentence.
In common with many other researchers, we believe that the importation theory and the deprivation theory are complementary in explaining the prison experience. Our research, however, focuses not only on the prison experiences of women, but also on the ways in which these women 'survive' their sentence. Given the fact that most women survive their detention and come out of prison alive, the term 'survival' should not be taken literally, but this is not to say that the female prisoners we interviewed all came away 'undamaged': many women in detention reportedly suffer from depression, posttraumatic stress disorder and other psychological complaints [6,25,26]. In the following, the term 'survival' is used to mean 'staying sane'.
Previous studies on women's prison experiences [3][4][5][6]10] have demonstrated that control over privacy is a necessary condition for women to endure a prison sentence. When female detainees are unable to establish a sufficient level of privacy within their immediate environment, such as their own cell, this may lead to self-destructive behaviour [10], although there is also some evidence that too much privacy leads to self-harm and suicide ideation (see [5,27]). Besides concerns about privacy, a meaningful daily routine can also help to make a prison sentence more bearable and improve social relations within the prison. When women are imprisoned, their main concern is about their lack of contact with family members and friends and the fact that they are no longer able to take care of their families [4]. Women seem to survive a prison sentence by constructing a life within the prison walls where privacy and safety are guaranteed. A relatively interesting daily schedule and a reasonable level of comfort are also important, but crucial to the survival of women in prison appear to be their relationships with fellow detainees, to the extent that they provide a sense of security and help prevent destructive behaviour [7,10]. This brings us to what Lazarus calls coping strategies. Coping has to do with the ways in which an individual handles stress and stress-inducing situations [28,29]. When a situation is appraised as threatening (primary appraisal), people evaluate their resources and options for responding to this situation (secondary appraisal) and search for appropriate coping strategies [24]. Examples of such coping strategies are: making plans to change stressful situations, accepting responsibility, denial or distancing, daydreaming, seeking support and comfort from others, positive thinking by emphasizing personal growth, social isolation or withdrawal in grief, stress reduction through drinking, smoking, drug use, playing sports, or through humour. Every person will use different strategies to deal with difficult situations and/or employ multiple strategies simultaneously [30].
As mentioned earlier, the manner in which prisoners deal with their sentence is determined by their particular life experiences as well as by the nature of the circumstances of their detention. Their choice of a particular coping strategy is likewise determined by importation and deprivation variables [24]. There appears to be a striking continuity between the coping strategies used by prisoners in their life prior to their detention and the coping strategies they resort to in prison [22]. Two prisoners can react totally differently to a lack of family contact. Whereas one prisoner, as a result of his or her character traits and experience of life, may seek out new social contacts examine the conditions and hardships the women face in prison, as well as their ways of coping with these conditions. We will try to show how the coping strategies they use are possibly connected to their life experiences and to prison life in Peru. The results could perhaps be generalized to women from different origins, though it would depend on the similarity of the country they came from. Janssen [16] showed in her research that women from different countries, experience their time in prison in The Netherlands in different ways, due to culture and other importation variables. Furthermore, it should be noted that due to the small sample size, it may be difficult to draw conclusions that are generalizable to the larger population.

Method
Ethnographic research methods can yield rich and vivid descriptions of people in their everyday surroundings, especially when the researcher is prepared to put up with a certain amount of discomfort [37][38][39]. This is not to say that researchers should identify with their informants, but rather that they must be willing to participate as much as possible in the world of their research subjects, even though this is not always feasible in prison research [38,39]. In our research, we have tried to give as detailed a picture as possible of Peruvian prisons and the experiences of their Dutch inmates with the aim of contributing to the development of theories on coping in detention. By describing both prison life and our research experiences as realistically as possible, we hope to add to the existing body of knowledge in this field. It should be noted that ethnography involves more than just telling 'interesting stories'. The rich descriptions and analyses derived from this method constitute an important tool for answering research questions and thereby contribute to the further development of more general theories [38,39]. On a more practical level the ethnographic method helped us to gain the trust of our informants. When we first visited the prison, most women were reluctant to talk to us about their lives because they found it too painful to revisit the past, but they soon got used to our recurring presence and this allowed us to gradually gain the inmates' trust.

The field
Peru has 66 prisons, run by the National Penitentiary Institute (INPE) and the National Police of Peru (PNP). In 2012 53.203 people were held in these prisons, of whom 3241 were women. Overpopulation existed as the prisons were built to keep 28.508 persons only [1]. According to Peruvian government statistics from 2012, the penal Santa Monica housed 1285 women, while it was only designed to accommodate 738 inmates [40]. It goes without saying that such overcrowding has a major impact on the life of all prisoners.
In 89% of the facilities, water and electricity were in fair or poor condition. Medical treatment services were said to be fair or poor in 71% of the facilities [41]. 32 of the prisons for women, concern mixed gender prisons. The mixed prison in Huaraz where we conducted our research, for example, accommodated 42 women opposed to 548 men. As these prison facilities were originally built for men only, women often cannot use the services available for men, because they are held in a separate wing. This affects their conditions of detention (Ministerio de Justicia & [42]).

Access to the field
We began our research at the foreign desk of the Dutch Probation Service (Bureau Buitenland van Reclassering Nederland), where we examined 325 files containing information on women incarcerated abroad. These files consist of a form, filled out by the detained women, about their background. Also reports on how the women are doing, written by a volunteer who visits the girls every month, are part of the file. Beforehand, we had established a set of criteria to identify relevant cases. The first criterion concerned country of birth: the women had to have been born in the Netherlands because the larger research project of which this article forms part also looked at the drug networks of Dutch women. The idea was that these networks would be different for women born in the Netherlands than for Dutch passport holders from the Netherlands Antilles. The second criterion was that the women had personally smuggled cocaine, as the larger project was originally designed to examine the motives of cocaine smugglers. On the basis of these criteria we compiled a list of all Dutch female cocaine smugglers detained abroad. We then selected the countries where eight or more Dutch women were being detained and read the files of women suspected of smuggling cocaine who were born in the Netherlands. In the end, Peru emerged as the most suitable place to conduct our research because it had the highest number of women detainees who met our criteria (i.e., fourteen in total). All women detained in Peru at time of research, were charged with drug smuggling. All of them left their homes for a 'short holiday break', not thinking about a possible confinement abroad. What remains, are children who need to be accommodated, houses that can no longer be kept, jobs or education that come to a hold, and family who suddenly has to take care of a family member who is imprisoned abroad (Ministerie [15]).
Access to the prisoners was facilitated by the foreign desk of the Probation Service. Its officers maintain close contact with the Dutch embassy in Peru, which is able to arrange visits to local prisons. Their assistance made it relatively easy to gain entrance to the two Peruvian prisons where Dutch women were detained: the Penal Santa Monica in Lima and the Penal Huaraz in the mountain town of Huaraz. We also visited and interviewed a number of women who had already been released on probation ('semi-libertad'). In Peru, prisoners who demonstrate good behaviour are released after serving two-thirds of their sentence, on condition that they remain in the country. We were able to find some of these women because one of them had become friends with a female guard at the gate of Santa Monica. She became our 'gatekeeper' and helped us get in touch with other Dutch ex-detainees.

Data collection
The research data were collected in April and May of 2010. At the time, there was still no treaty between the Netherlands and Peru allowing the transfer of prisoners. 2 This meant that Dutch nationals arrested and convicted in Peru had to serve out their entire sentence in a Peruvian prison. Several of the fourteen women on our list were not willing to participate in the research. It is unknown why, as the other respondents simply told us that these women did not want to meet us. Unfortunately we could not verify this, as the women unwilling to participate, stayed in their cells. The researchers were not allowed to enter the cell blocks, and depended on the women who were curious about our visit, and were eager to come and talk to us. We were also unable to track down all of the women who had been released on probation.
In the end, we were able to interview and spend time with ten Dutch women convicted of smuggling cocaine, both in and outside the prisons. Five of them were still in detention, while the other five were living outside the prison walls. Although our research was primarily focused on prison experiences, we also interviewed the women in semi-libertad, as they had been released from prison in the last year and were still under the obligation to report regularly to the prison authorities. We asked the women on probation about their time in prison in retrospect; one would expect that these women look back differently on their experiences in prison once they are released. However, in this study there were no differences found between the women who had semi-libertad and the women who were still incarcerated, perhaps because the women on probation were still Bdoing time^outside the prison walls.
The average age of the women we interviewed was 35. The youngest was 18 when she was arrested, the oldest 54. Five of the women were mothers. Half of them had previously been addicted to alcohol or drugs, and half had had prior police contact in the Netherlands, ranging from fines to serious criminal records. All of the women had a low level of vocational or secondary education and most of them had accumulated sizable debts. Two informants had served time in foreign prison before.
We conducted in-depth interviews because an open-ended method of data collection seemed best suited to answer our research questions. Guided by a list of topics, we asked the women about their life stories. These topics included (but are not confined to) educational and working careers, social contacts, criminal histories and prison experiences. Each respondent gave us her verbal consent to participate in the research.
An additional advantage of conducting interviews was our physical presence in the field. This enabled us to observe prison life firsthand and experience for ourselves the conditions our respondents were faced with on a daily basis. For example, one of the prison guards offered us the opportunity to jump the queue at the prison gate for the equivalent of EUR 2,50. We were also able to experience what it meant to have to share a small space with an ever-growing population of inmates. The interviews in prison were mostly conducted on visiting days, which meant trying to have a conversation in an overcrowded prison courtyard.
We met most of our respondents two times. The first meeting was meant to introduce ourselves and to plan an interview. The second meeting lasted in general 1.5 h, in which the interview was conducted. We met three of the women who were on probation again after the interview, either to go for a coffee, to exchange books, or just to break the daily routine of their lives outside of prison.
All interviews conducted outside the prison were recorded with the help of a voice recorder and later transcribed verbatim. Unfortunately, we were not able to bring recording equipment into the prisons. Despite our many requests to the guards, to the female warden of the women's prison of Santa Monica, and even to the president of all Peruvian prisons, we never received permission to record our conversations with prisoners. In a bar across the street someone offered to smuggle a recorder into the prison through a corrupt prison guard, but we declined the offer so as not to endanger the project. Instead, we wrote down the women's responses as fully as possible and transcribed our notes immediately after the interviews to retain maximum information. We used pseudonyms to ensure the anonymity of the participants in this study.

Data analysis
For analysis, the interview transcripts were coded by two researchers. By reading the interviews and labeling any fragment, sentence or paragraph within an interview that seemed to represent topics that were of interest, a first analysis was done. The researchers compared their codes to obtain researcher triangulation. Unfortunately no inter-rater reliability was measured, which could be a potential pitfall of this study.
What helped the researchers focus, were the research questions and topic lists that were based on literature about characteristics of mulas [16,43], coping [30], and prison experiences [4,44]. During the interviews the researchers asked about these themes and topics, for example about their lives before the imprisonment and about their lives within the prison walls; these themes and topics could be labeled as such during the coding. After coding the interviews, all labels on the same topic were put together in one document, so similarities and differences between interviews on a certain topic, could be easily found.

Life experiences of Dutch mulas 3
The Dutch women who are imprisoned in Peru, experienced a relatively large amount of adversities in their lives: violent relationships, addiction (personal or a family member), and mental health problems like depression and burnout were common. Most of them had problems when they were still in the Netherlands, and some could no longer work due to psychological reasons. It seems that these women smuggled drugs out of desperation. For example Louise, a 56-year-old woman, who had to stop working due to a burnout, and had been under the scrutiny of debt restructuring for 7 years before she was offered 5000 euros for smuggling drugs at a birthday party.
Nearly all the women we spoke to had left their parental home at a very young age; some of them moved to boarding schools, while others moved in with their boyfriends. Most left their parents' house due to problems at home or because of the (temporary) incapacity of the parent or parents to take care of them. For instance, Nicky left her home at a very early age; at fifteen she ran away and moved in with her older and violent boyfriend. She was led into a world of drugs, which her boyfriend was already acquainted with. From that moment on, she dealt drugs from her scooter, and refused to obey the law. It appears that she lived on the edge when she was still outside of prison, satisfied with earning money through drug dealing and her search for adventure. Moreover, it seems that Nicky did not conform to the rules in prison either. She withdrew from taller (manual work), and resorted to drug use. This seems to confirm that coping strategies are indeed influenced by former life experiences as Nicky again did not obey to any rules, though her coping in prison may not have been a rational choice but her only option to survive.
There are more examples of women who appear to use coping strategies determined by former life experiences. For instance, Karin, who seemed to be an aspiring and welleducated woman before her imprisonment. Behind prison walls she became the spokesperson of many fellow inmates, and helped them to fill out their forms. Not only did she do this to help those who were in need, Karin also assisted other women to make herself feel useful. This reinforced her self-esteem, which helped her to survive the difficult time in a Peruvian prison.
Eva was a quiet, but also depressed woman prior to her detention, who struggled to keep her head up during her detention. In prison she preferred to withdraw to her cell. The ways in which these women handle their detention abroad, and the coping strategies they adopt, mainly seem to be determined by their course of life.

Deprivation in a Peruvian prison
The circumstances in Peruvian prisons seem to be tougher than circumstances in detention facilities in their home country. According to the interviewees, corruption is for example an unfortunate part of daily life in the Santa Monica and Huaraz penitentiaries. Prison guards earn such a small amount of money that it makes it easy for prisoners to bribe them. But the guards would not keep the money. A major part of the corrupt income is said to be handed over to the local prison director, who in turn divides the money among the staff. Corruption, so it seems, is a means used by prison personnel to gain extra income.
Among the detainees there is not a lot of physical violence, and according to the women interviewed, the prison guards do not use any force on them either. The exclusively female guards are not allowed to touch prisoners and there are a remarkably small number of prison guards present within the prison compound. Approximately 12 female guards watch a group of roughly 1000 women. Delegadas, who are representatives of the detainees, keep order. These delegadas make the rules on a piso [ward] and are, for instance, responsible for the shared finances of cleaning goods. Due to a shortage of prison staff, the Peruvian prisons are actually run by prisoners. According to Peruvian law, this is prohibited: convicts are not allowed to be in a dominant position over other convicts. However, the shortage in personnel is solved this way. To prevent a breakout, the prison walls are guarded by heavily armed men.
Rules, set by the prison regime, seem to be arbitrary. One warder allows certain things while another forbids the same, and rules can change overnight. This causes a lot of frustration among the imprisoned women. We as researchers were also confronted with the arbitrary rules: 1 day a letter from the Dutch embassy was sufficient to enter the prison, but the next day we were denied entrance because the same letter was rejected.
Besides the difficulties named above, there are certain things that many of the Dutch women in Peruvian prisons find tough. Though the imprisoned Dutch women each experience their prison time in their own way, there are some common aspects, like the ones named above, that most of the women find hard. The mentality of their Peruvian fellow inmates is strange to them. Dutch women find them impudent and unhygienic and they are annoyed by their manner of talking, which is more like screaming. Also, the Dutch detainees we interviewed reported that they detest all the gossiping that would happen because of jealousy. Some women see the Peruvians as the initiators of gossip, others seem to think that Dutch women gossip as well. It is unclear though, whether the Dutch women had a sufficient knowledge of the Spanish language to perceive gossiping. As well as speaking badly about each other, the women reported that they do not have a lot of respect for one another's belongings. According to our informants, stealing among Peruvian women would occur regularly.
The women's prison in Lima is overcrowded, which makes it difficult to avoid people or to withdraw from the crowds. Also, hygiene is inadequate. It is very nasty in here, Shirley says, because you don't put your toilet paper in the toilet but in the basket next to it. 150 women, four toilets, no way I'm going to clean that. And sometimes there is no running water. The other day there was no water for two whole days! But I just pay 20 cents so I can use the toilet upstairs at the sickbay. But even upstairs there's no way they leave you be, banging on the door because they want to use the toilet too (Shirley, 38).
The lack of privacy is considered to be the toughest thing about their stay in Santa Monica prison. The cacophonic sounds, caused by the presence of thousands of women, make the mulas realize that they are never alone. Michelle proudly recalls memories of the curtains she had made to give her some privacy in her upper bunk bed. I had a curtain made around my bed, which I could close, 'ha, alone!' [laughs]. 'Cos when I slept in the salon [TV room] I wasn't allowed to have a curtain 'cos then no one could watch the TV. So yes, sometimes I would sit with someone who had a lower bed or had her bed in a cell, but you'd never have real privacy. Only after I got into that cell room, after a year, I had my curtain. You could just close your curtain and when someone comes in to see you, you say 'I'm not there!' (Michelle, 21).
As well as the penitentiaries in Lima being overcrowded, medical care within the facilities is said to be seriously lacking. Some of the women have seen fellow inmates die and feel that this was unnecessary. If you didn't have the money they would just leave you to die. Everything costs money. So if you don't have cash, you can't go to hospital. You know, people with AIDS in there died, only because they didn't get medical treatment. I've seen that a few times, people who died because of a lack of medication (Karin, 36).
The geographical distance between Peru and the Netherlands is large and it is hard on the women that because of the long and expensive journey they receive very few visitors. The long distance between them and their families is perceived to be very hard: for example, to know that family on the other side of the world are not doing well is seen as extremely tough. Also, the limited contact moments are difficult for them. Telephones can only be reached after a very long queue. When it is an inmate's turn to use the phone, usage is limited to 10 min only. Letters and postal packages are brought in once a month by the Dutch embassy or volunteers, and every Thursday, during Bible class, a maximum of three e-mails are allowed to be sent.

Survival through persistence
Documentaries that can be seen on television often show the excessive violence that exists in South American prisons. This is the same image that most Dutch women have when they enter a Peruvian prison: 'I was so scared (…) a nightmare that they would rape me, that they would beat me (…) you just don't know. The only thing you know is what you have seen on TV' (Agnes, 41).
But the women we spoke to compare Peruvian female prisons with boarding schools instead. The threatening image they had of South American prisons did not come true. On the other hand, the women did find life in these 'boarding schools' hard. 'But you have to go on' is what they frequently told us. How do these women survive their sentence in a foreign prison? What coping strategies do they use and do these strategies correspond to the strategies that were found in existing literature?
During our field study we found the following coping strategies: (conversion to) religion, bonding through play families, bonding through romantic relationships, join in activities, help fellow inmates, refrain from thinking, and put the own identity on hold. It seems that the coping strategies that are used by Dutch women to survive their detention in Peru, are determined by their experiences in life and by the prison context. For example, six out of ten Dutch women use their faith in God to stay strong. Religion is a coping strategy that is also named in existing literature. Two women were already religious when they were incarcerated (import), but the other four women got to know God while in prison (deprivation). Every Thursday afternoon Mary comes by. She is an American reverend's spouse who teaches Bible study. Most Dutch women participate in these classes. They explain that their faith in God helps them to keep going, even if they are in prison far away from their families.
We found that Dutch women that are imprisoned in Peru, cope with their imprisonment by mutually bonding with their fellow inmates and forming family ties. A fellow inmate then becomes a sister, a mother or a niece and with them they share their joys and sorrows. This can also be seen as a coping strategy: it is a response to prison circumstances and them missing their own family. The women imprisoned in Peru also use this coping strategy to survive: 'You kind of observe who you can trust and then you sort of start a relationship. If one didn't have money, the other one had. You took care of each other. Also emotionally: if you got bad news the other one was there to support you' (Karin, 36).
Furthermore, six out of ten women engaged in romantic relationships inside or outside the prison. Their need for intimacy and security is, for example, filled by a romantic relationship with a fellow inmate. Sexual relationships are a major taboo in Peruvian prisons, but still many detainees choose this form of intimacy. Though almost every Dutch prisoner denied being in a relationship with another woman, they pointed out that at least half of all the female prisoners started a relationship with one of their fellow inmates. 'At a certain point you just have a need for intimacy,' they said. As well as lesbian relationships, affinities between male and female detainees also occur. The women can write to men who are detained somewhere else in Peru or they meet each other while being transported to court. Shirley corresponds with several men. She giggles like a schoolgirl when a letter arrives. 'Well, you have to go on, on with your life, even if you're in prison,' she says.
Typical of prison life is the battle against time. Women in Peruvian prisons try to keep busy so that they 'won't go out of their minds'. The ways in which they try to keep busy differ. Some of them focus on taller, others join in sports tournaments that are organized in prison, but most remarkable is that some of the women mainly try to help others. Not every woman speaks Spanish, especially not the newcomers, so some women help others to fill out papers or translate important documents. Not only does this help their fellow inmates, it also enhances their self-esteem, which contributes to their struggle to survive their prison sentence. It is not a coincidence that some women use this coping strategy: they happen to be social workers outside the prison and so this coping strategy is a derivative from former life experiences. The circumstances in prison ensure that others actually need their help: for example, the lack of interpreters makes their help extremely necessary and welcome. Karin (36) claims to have helped many fellow inmates when she was still detained in Santa Monica prison: Everything happens for a reason, I think. I've helped tons of people inside, with papers, translations and so on. So maybe it was for a reason that I had to stay that long. For me it was also some sort of survival. I always worked in community serving jobs and for me it was a way to feel better. It made me feel better when I was still in prison. And it kept me busy.
A different survival strategy is to stop thinking. By refraining from thinking about the imprisonment and its consequences, the women try to cope with their incarceration. Jelena (24), for example stated: Just don't think, that's the secret. And don't have too much pity on yourself. The people I love, they're simply not here. But if I keep thinking about home and about my baby I won't make it. This is where I live now, and this is where I need to succeed. Some people are complaining all day: 'I can't do this anymore, I can't take it any longer.' And of course you're allowed to complain sometimes… but not all the time. 'Cos then I think fuck off, I have to do 14 years and you'll be out in just 2. But I keep smiling. If I feel depressed, I'll just stay in bed and afterwards I'll come downstairs like nothing happened. I always smile because otherwise I simply won't make it.
The circumstances in prison may also lead to the inability for women to be themselves, in order not to show any weaknesses. Some of the women in Peru put their true identities on hold, in order not to show any weaknesses. Eva (31), who was being treated for depression in the Netherlands prior to leaving for Peru, puts it like this: My depression is still there but I try to push it away. 'Cos if you show weakness in here, they will also treat you like that. So I think I just live on autopilot.
Furthermore, the length of the prison sentence influences the coping strategies of the female detainees we interviewed. Women who smuggled less than 10 kg of cocaine, were sentenced with 6.8 years in prison, with a chance on sentence reduction, in case of good behavior and joining taller. Women who smuggled over 10 kg of drugs, were sentenced with 15 years in prison, without a chance on reduced penalties. This last group of long-term prisoners deals with many of the same stressors as the women that 'only' got 6.8 years. But the women with shorter sentences are able to work towards a goal: sentence reduction. This gives the opportunity to use good behavior as a coping mechanism, to leave the Peruvian prison sooner. Women with a long-term sentence on the other hand, have little to gain from good behavior and taller, and many times withdraw themselves from both, fleeing for example into drug use.
Dutch women experience and survive a prison term far away from home after being arrested for drug trafficking.
In the first place, our research showed that the lives these women lived prior to their detention were often distressful. Furthermore, the circumstances they find in prison seem to influence the way they experience their imprisonment and the way they survive their prison term.
Prison life in Peru is very different from life in a Dutch prison and Dutch inmates in Peru are faced with all sorts of problems that are usually non-existent in their own country. The lack of privacy and the arbitrary regime in the Peruvian prison are tough on these women. According to the interviewees, health problems may originate and worsen due to a lack of hygiene and medical care. Furthermore, a language barrier often exists and most Dutch prisoners receive no visits from relatives. The circumstances in Peruvian prisons therefore seem to be tougher than circumstances in detention facilities in their home country. Because of that, the women in this small-scale study may experience harder deprivations than women in existing research on prison experiences.
Earlier research shows that prison conditions are of major influence on the wellbeing of prisoners [26,45,46]. Comparative research between prisons in California and The Netherlands, for instance, showed that female prisoners in The Netherlands were less depressed due to the relative humane circumstances in Dutch prisons [45]. BWhere he or she sleeps, what and where he or she eats, whether he or she has a bed with sheets and blankets or sleeps on the floor, covered only with rags, all this has tremendous influence on his or her physical and mental well-being^( [45], 11). Comparative research between prisons in the United States, England and The Netherlands also showed that the particular nature of the prison regime, in the end explains the differences in the prisoners' psychological well-being, regardless of where they were imprisoned [46]. So we can imagine that serving time in Peru, is different than being imprisoned in Brelatively humane Dutch prisons^( [26], 191), due to the differences in prison conditions. As no comparative research has been done between Peru and the Netherlands yet, no sure claims can be made. But as Kruttschnitt et al. [46] and Slotboom et al. [26] showed, the prison conditions determine the psychological wellbeing and the experienced amount of stress. We suggest that the amount of stress determines whether coping strategies are needed or not. It therefore seems plausible that the need of coping strategies is higher in Peruvian jails due to the amount of stress.
Though a need for coping strategies might be higher in Peru due to the stress caused by deprivations, it is carefully suggested that the ways in which our informants handle these deprivations do not differ from the coping strategies that were already found in prior prison research. Our respondents found support in religion and surrogate families, tried to help other inmates who are struggling, or simply tried to think as little as possible about their situation, similar to what earlier research by Clear et al. [25], Lazarus and Folkman [30], and Owen [10] showed. The coping strategies that the Dutch women used can possibly be traced back to the importation and deprivation theories discussed in the literature. The choice of a certain coping strategy seems to depend on prior life experiences and the circumstances in prison. With that, the coping strategies found in this small-scale ethnographic study seem to correspond with the general coping strategies described by Zamble & Porporino [24].
A drawback in this study is the small number of respondents: it is therefore difficult to draw conclusions on this small-scale study. It would be fruitful to conduct more interviews with women who are imprisoned abroad to be able to further compare whether cultural background and prison conditions influence the experiences of incarcerated women.
We conclude to state that, though the imprisoned Dutch women lived tough lives and their stay in a foreign prison is hard, they manage, just like detainees who do time in their own countries, to cope and to survive.
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.

Introduction
In Italy there are seven women's prisons and 72 female divisions located within the male prisons. Although the number of female prisoners is increasing, they are still a minority compared to male prisoners. Official data updated to the year 2015 (31st March) illustrate the presence of 54,122 prisoners in Italian prisons, 2354 are women (4.35 %), half of them are foreign prisoners. Only a small number of the women are married, the others are single; however, about 50 % of these female prisoners have children.
The recidivism rate is very high, especially among women detained for drug-related offences, the most common crime after property offences. Most of the inmates serve mild sentences, not over 5 years, and there is a high turnover rate due to the short sentences.
The specificity characterizing women prisoners in Italy is the type of crime: theft, drug trafficking and exploitation of prostitution appear to be the most recurrent. In particular, data from the Ministry of Justice, year 2012, demonstrate that: 1. The rate of women incarcerated for drug-related offences is 33 %; these women are young and the rate of female incarceration is higher than that of men detained for the same offence. The women prisoners for drug trafficking are mostly all foreign, while the drug-users are mostly Italian. 2. 22 % committed property offences; these crimes were committed mostly by young women who wanted to achieve economic autonomy away from the family or women who had to support their young children without a partner's help and the support of social services. 3. 12 % committed crimes against people, this is a low rate compared to males who committed similar offences. 4. 50 % of the female prisoners have children but have interrupted all relations with them, and many, in particular, the Romani women, have dependent children under the age of three living in prison with them. Over the years, new laws have been introduced in Italy which allow mothers and children to live their detention in group homes or in their own houses.
Foreign female prisoners, a minority in Italian prisons, face many factors which make prison life more difficult. The critical conditions are in fact represented by additional factors, such as: 1. it is difficult for a female immigrant, without a house, to have access to an alternative to prison; 2. the condition of clandestine family or relatives minimizes the possibility of having visits; 3. it is difficult to keep telephone contacts with their families; 4. difficulty in raising their children.
The limited number of women's prisons makes the presence of female sections in male prisons necessary; however, this situation often makes women prisoners feel forgotten, especially by intramural educational programs targeted for the average male. In fact, these limited numbers do not allow the realization of projects related to educational or professional courses or, in any case, projects aimed specifically at "women". Being a minority within a minority, being equal but different, being forgotten because not important for statistics: this is the condition that women are forced to face along with their sentence, living in a universe conceived and organized on the needs and characteristics of the average male prisoner, as we will consider infra.
This situation also occurs to women prisoners with health problems. They live a condition that I defined "double burden" [1]. In fact, international studies demonstrate, through empirical investigation, not only that the number of women in prison compared to men is increasing significantly [2], but also that for women the presence of disease is often higher than for men; this includes mental illnesses [3], drug addiction [4,5] and sexually transmitted diseases [6,7]. Moreover, women are not only more exposed to certain types of diseases, but are often affected negatively by intervention and therapeutic programs created specifically for men [8], therefore making the period in prison less tolerable for women than for men [9].
An interesting Report of the World Health Organization [10] stresses the lack of public health concern for the needs of women in prison considering the following four main issues: In particular, the mental health problems "should be comparable with those available in the community, but should take into account the negative effect that prison is likely to have on the mental state of a woman" [ivi: 158].
Regarding the second point, data show that women are 14 times more likely than men to injure themselves while in prison, and that the early period in custody is recognized as the high-risk period for suicide.
Problems linked to substance use are principally determined by the fact that women often neglect their health while free due to poverty and/or addiction behaviours, with the result that after entering prison they make great demands on health services [ivi: 161].
Finally, pregnancy and motherhood -as we will see infra -have a sensible effect on the health and well-being of this sample of people, not only from a bio-medical point of view but particularly at a psycho-social and relational level, as Enos asserted [11].

Theoretical framework
Well-being may be considered in two perspectives: the "hedonic" approach, which focuses on happiness and defines it in terms of pleasure attainment and pain avoidance; and the "eudaimonic" approach, which focuses on self-realization and defines wellbeing in terms of the degree to which a person is fully functioning [(12): 143-147]. Well-being is strongly linked to the health status in both cases: in fact, "sickness is often associated with displeasure or pain, so the presence of illness might directly increase negative effects. Furthermore, illness often presents functional limitations hindering positive effects and life satisfaction" [Ivi: 151].
The theoretical model of this article is the Self-determination Theory, that is under the umbrella of the "eudaimonic approach". This Theory describes three main needs: autonomy, competence, and relatedness. This study is focused especially on the relatedness need. The model helps us to build our research hypothesis: stress in women prisoners is strictly related to both "external" networks and "internal" support. Furthermore, it is generated by the feeling of loneliness rather than having to cope with the difficulties of prison life. The main issues of this article concern stress and loneliness that women prisoners experience in Italian prisons.
In general, wellness is not simply defined as the absence of psychopathology, but as an array of positive functioning aspects promoted by attainment of relations of strong attachment, acquisition of age-appropriate cognitive, interpersonal and coping skills, and exposure to environments that empower the person, as discussed by Cowen [13].
A significant indicator of well-being, especially from an "eudaimonic" point of view, is relatedness: quality and quantity of relations are fundamental to achieve wellbeing outcomes. Relatedness is one of the most significant factors that influences happiness [14]. Specifically, the quality of relatedness is a more important predictor of well-being than quantity, in fact, people who have more intimate relations tend to perceive greater well-being. Reis et al. [15] showed that people experienced greater relatedness when they felt understood, engaged in meaningful dialogue, or enjoyed themselves with others.
Social support and positive relations also have an important effect on individual health: as Uchino et al. [16] asserted, social support influences mortality via changes in the cardiovascular, endocrine and autoimmune system. According to the "eudaimonic" approach, another indicator of well-being is selfefficacy: therefore, on average, goal progress predicts enhanced well-being, in particular goals that are rated as important [17].
Scientific literature defines stress as "any environmental, social or internal demand which requires the individual to readjust his or her usual behavioural patterns" [18]. Peculiarities of total institutions, which constitute a real fracture with the prisoner's precedent life, can surely be an obstacle for the health and well-being of these people.
In the prison environment, it is possible to differentiate first-level stress (or the "primary stressor"), due directly to the "internal" experience of incarceration, and second-level stress (or "secondary stressor"), due to the "external" experience (post-release life and return to the free world) of social stigma, reduced job opportunities and problems in the family [19]. As a primary stressor, life in prison requires the inmate's capability of "adjustment" and ability to cope from the first moment and in relatively short time periods, something that rarely happens to people who live what we called "double burden".
More specifically, loneliness is defined as "the unpleasant experience that occurs when a person's network of social relations is deficient in some significant way, either quantitatively or qualitatively. Loneliness can be mild and fleeting but it can also be a persisting, distressing experience" [20]. Since loneliness is linked to a lack of interpersonal relations, it should also be related to measures of physical health.
Weiss [21] described two forms of loneliness: loneliness as social isolation and loneliness as emotional isolation. The former results from an absence of supportive social networks, the latter stems from the absence or loss of close attachment relations [22]. Such loss of relations "certainly occurs with prisoners. Both solitude and social interaction are self-chosen states that prisoners can only experience in a restricted manner. Because of this limitation, prisoners may be even more vulnerable to loneliness" [(23): 6].
Loneliness is also an important indicator of suicide in prisons: "isolation from family members and outside contacts, as well as not being able to make new friends in prison, for example, are loneliness related factors which potentially make a prisoner more vulnerable to suicide. For many prisoners incarceration is where relations are interrupted and bad news is not uncommon. These situational triggers may evoke a sense of loneliness in some prisoners" [(24): 435].
When people cannot satisfy this need, they experience social loneliness, a feeling of marginalization and boredom. 2. Emotional needs. It refers to the attachment bond that individuals activate first with parents and then with a partner when mature. If not fulfilled, feelings of emotional loneliness, insecurity, sense of abandonment and lack of protection may appear. 3. Sexual needs. That is sense of intimacy and body contact. It is associated with attraction and desire. If not fulfilled, individuals may live experiences of sexual frustration and sexual dissatisfaction.
Relatedness, social support and emphatic relations are fundamental to explain the well-being of women inmates. According to Lopez's categorization [Ibidem], social needs are specially investigated to understand the sense of boredom and seclusion of the prisoners.

Literature review
In prisons, many needs should be satisfied to avoid loneliness and seclusion. Toch [29] identified seven environmental needs for prisoners: 1. Privacy. Some prisoners may choose to escape from crowds and noisy settings. 2. Safety. Inmates may prefer a safe environment to minimize chances of being attacked. 3. Certainty. Prisoners may express needs for clear and consistent rules and procedures. 4. Assistance. Inmates may ask for help to solve practical problems or to obtain services. 5. Support. Prisoners may have needs for understanding, empathy, warmth, emotional support. 6. Activity. Prisoners need to be occupied. 7. Autonomy. Prisoners should be able to control their lives, minimizing restraint.
Prisoners may obtain social support from two main groups: outside the prison from family, professionals and friends; inside the prison from staff and fellow inmates. If the relations outside prison may be informal, in prison they become formal and top-down directed.
Compared to Goffman's intuition of total institutions [30], contemporary studies show a different picture: an interesting empirical research [31] shows a new role perception among prison officers. Many prison officers have pro-inmate orientations and strive to encourage inmate rehabilitation while Fuller suggests that prison officers now encourage a greater level of effort, support, and openness and are interested in expanding the supportive aspects of their duties [32].
This support may be considered as [( Regarding relations with other fellow inmates, many researches show that prisoners are not a critical core, but rather atomistic and distressed, with no determination in activating deep relations. This is explained by Zamble & Porporino [34], who argue that inmates are cautious about opening themselves to possible entanglements for fear of getting involved in fights in order to help another inmate. What is more, Biggam & Power argued that "the nature of the prison regimewith limited hours for social contact in privacyhinders the formation of close friendship" [(35): 226].
Studies on stressors in the prison environment revealed common factors for both men and women. These are loss of freedom, lack of opportunities for heterosexual activities, no support from family and friends, depersonalizing experiences, loss of autonomy, lack of privacy and security [36].
The most significant stressor for female inmates is linked to their role as mothers: the separation from their children. This stressor is often associated (see infra) with feelings of guilt, anxiety and fear of losing mother-child attachment [37]. In a longitudinal study, Fogel & Martin [38] compared anxiety and depression longitudinally between mothers and non-mothers in prison. The findings showed that although anxiety levels decreased over the length of incarceration for both groups, it decreased much more for non-mothers than it did for mothers.
Female inmates are more distressed than men as they have to cope with more stressful events, such as the separation from their children [39], and having to find someone to take care of their children. This situation "is stressful for mothers because of the possibility of losing the custody of their children added to the worries associated with the well-being, education, and raising of their children" [(25): 646].
Studies on loneliness in female prisoners show heterogeneous findings. Ann Desmond's research does not support the hypothesis that the frequency of family and friends' visits decreases loneliness. On the contrary, the research indicates that "the presence or absence of a friend and the number of friends within the prison did make a significant difference in the loneliness experienced by women prisoners. Women with a friend in prison were less lonely. Women who had more friends were significantly less lonely than those who had few friends. It also appears that women are more likely to make friends in prison than men" [(23): 9]. Moreover, women show higher participation rates in groups and clubs in prison, and "their lives seem to be organized around small, intimate and make-believe families" [(25): 650]. Therefore, according to Desmond, it seems that having external social support did not decrease loneliness, while having internal social support was associated with lower levels of loneliness. Contrarily, men are more focused on making time pass and solving their problems by themselves. "Men do not search for support as much as women do inside prison, which implies that men inmates receive less support and therefore may suffer more social loneliness" [Ibidem].
Regarding relations with the outside world, Lindquist [40] found that being married, being a parent and receiving social support within the prison were all associated with poor or low mental health. Therefore, the authoress concluded that social integration is negatively correlated with mental well-being. Specifically, married inmates reported higher levels of depression and anxiety, and inmates with closer social relations inside prison reported higher levels of hostility.
Schmid & Jones [41] suggest a way of adapting to prison life: reducing contact with the outside world because these ties make "doing time" more difficult. Inmates are tempted to discontinue outside contacts, which may become painful reminders of what they left behind upon entering prison. Contrarily, Biggam & Power [35] found that the discrepancy between "expected" support and "received" support may predict mental health: emotional discrepancy may be the best predictor of anxiety, depression, and hopelessness.

Aim and hypothesis
The aim of the article is to investigate, through an empirical survey, the health and wellbeing of women inmates in Italian prisons. An important issue is the loneliness and seclusion that these inmates experience inside prisons.
The hypothesis is that stress in women prisoners is strictly related both to "external" networks and "internal" support, furthermore, it is generated by the feeling of loneliness rather than having to cope with the difficulties of prison life.

Methodology Participants
A qualitative study involving adult female prisoners in three Italian regions was conducted (Campania: prison of Benevento, Lazio: prison of Rebibbia-Rome, Emilia Romagna: prison of Dozza-Bologna) based on 37 individual semi-structured interviews.
The interviewees are mostly all Italian women, except for five, two from Africa, one from Albania, one from Romania and one from Spain. Their average age is over 40 (42.6) and the majority have motherhood in common (33 mothers were interviewed).
The sample is not representative in statistical terms of all female inmates in Italian prisons; nevertheless, it surely constitutes a reliable sociological sample from a qualitative methodological perspective.

Procedure
The qualitative research was based on semi-structured interviews, aimed at investigating mainly the experience of women prisoners in regards to living collectively and their well-being, but with the specific intent to explore new themes through their answers. In literature, life history narratives are considered as the optimal method to collect data from incarcerated women and other marginalized populations [42].
The interviews were recorded and then transcribed verbatim (word by word, without additions or omissions). The recordings were heard repeatedly before their transcription to grasp even the prosodic aspects of the narratives.

Analysis
Analysis was conducted through the CAQDAS contents (Computer Assisted/Aided Qualitative Data Analysis Software) by the software Atlas.ti. Using the Grounded Theory model [43,44], the text was analysed from the data and returning back to the text, in order to complete the construction of concepts and relations among them, letting the prisoners' words suggest the priorities. Accordingly, during the first stage of analysis there was not a plan ex ante but simply leaving recurrences emerge from the text in order to extrapolate one or more topics of investigation.
The contents were then initially encoded with an Open coding, creating categories mostly reporting the words of the participants; and only later were the "quotations" coded for recurrent themes in Code Families, in order to build families of concepts that could address reflections, also through the definition of relations within the Network themes. By crossbreeding the Family Codes, operative Super Codes were derived (by Boolean operators) which enabled to observe significant relations. This procedure allowed us to build reliable categories, in a bottom-up direction.
The role of the researcher is fundamental in this methodology; in fact, "while computer software can help with various stages of analysis, it will not perform analysis. The intellectual work of devising coding schemes and developing theory about the data is the responsibility of the researcher. Software is simply a tool that can help with the systematic sorting of data, if appropriately applied" [(45): 98].

Results
From the conversations with the prisoners, what mostly draws our attention is a concept, which, to summarize the complexity, could be defined as "emotional dimension". This concept especially regards the daily and personal experience of each detainee, and emerges in most of the conversations with the respondents: over 440 quotations (Citations within the text related to the concept) can be counted within this vast dimension, so articulated to offer the analysis a network of over 300 inter-connected concepts.
For example, it is possible to organize the network so that only the main connections emerge, which allow to understand this aspect in detail.
A bipartition of this Codes Family is further identified in the light of this inherent polarity in the dimension itself, which is split in the presence of a positive or a negative approach to incarceration. It concerns fundamental assumptions that recur throughout the investigation as sensitive issues regarding the life of imprisonment, according to the literature review above mentioned: 1. internal support by the specialized staff and/or by other prisoners, and external family support; 2. health and quality of medical care in prison; 3. activities to keep busy (projects, courses, work or training) and post-prison reintegration.
In particular, the prisoners without family affections, not finding a support network in the professional staff (due to staff shortage, little or no counselling with psychologists and social workers and mistrust of medical care) and solidarity with other prisoners, become dramatically demoralized. They also show severe forms of depression and even attempt to commit suicide.
Moreover, the availability of an external or internal psychological support is a decisive factor in the individual experience contributing to a positive re-elaboration of the prison experience, according to Toch [29].
The "positive emotional dimension" alludes to the individual disposition to live prison as a possible redemption (we find quotations as: "prison saved my life", "positive future", and so on). These prisoners take advantage of work opportunities and internal courses in the institution to build and plan their own future.
They also have total trust in the prison staff and have established good relations with the other prisoners. When interviewed they are positive about the opportunities they will encounter outside prison, but this is mainly due to the stable external support of their family. In fact, supportive social networks, together with close attachment relations, are fundamental to cope against loneliness, as we described in the theoretical framework in the re-examination of Weiss' typology of loneliness by DiTommaso & Spinner [22].
Here below are the answers to the question "How do you see your future?": "I'm not worried. I have a family waiting for me outside and supporting me, so I am calm. I believe that my future will be positive: my mother has put some money aside for me and I think I will open a business, so I can work and have a life like everyone else".
"I still believe in my future, even if I am 52 years old, I imagine it positive, I am confident, I want to believe that life continues for me, I will find a job, because here I have learned a lot from my mistakes, and now I want to start all over again. If I have any problems, I will find my children who are now grown-up and have their own life, if they have children I can take care of them, and we can start again together, I want to believe this".
On the contrary, the absence of support, which can derive from long distances from family members or due to distrust of prison staff, becomes the core of the "negative emotional dimension", in which loneliness is experienced as the real "illness" of prison life. In fact, higher levels of social loneliness are correlated with higher levels of poor mental health [46].
Without strong emotional support, the prisoner does not have the necessary motivation to deal with the trauma of incarceration. In these cases, the threat is not only in the future but also in the daily life of the present: imprisonment itself often becomes the "disease", especially at the first experience of imprisonment, as we discussed in relation to Pearlin's concept of the "primary stressor" [19], due directly to the management of everyday life in prison.
Here below are the answers to the question "Tell us about a typical day here in prison": "That's it, I fell into a deep depression, I was in bed every day, thinking, thinking again, then I had terrible headaches, dizziness, palpitations, I never slept, I was terrified of everything around me. I just did not want to stay here, I really felt bad, and I started like this here, taking anti depression drugs, I took so many, which, on one hand, calmed me down and did not make me think, but then, at the end of the therapy, as my inmates suggested, I started to do gymnastics, yoga, to attend courses in order to make time pass and enjoy myself a little. When I came to prison, it was the first time for me and I found myself alone, frightened, and then I also suffered a strong emotional stress, because they came to arrest me when I was breast feeding my baby, and I can never forget that scene, it was hard for me, I can still see my baby in my mind, and this has made me feel very bad here".
"You have to be careful here because we are not all strong: there are inmates who hurt themselves, how do you say it? self-harmers, I think, here, there are so many that it is also dangerous for those who are around them, they must be helped".

The experience of discomfort and never-ending fear
Studying the contents of the network analysis in depth, we can detect a new dimension, that is, the discomfort experienced by women in prisons. It is particularly intense both as an "existential discomfort", deriving from loneliness, from being away from home and the family, and as a "concrete discomfort", deriving from living their daily life in prison. Creating a Codes Family with the quotations from the conversations referable to this concept, over 340 quotations and over 130 links emerge, allowing to consider this theme central in regards to the actual analysis. Clearly connected to the negative emotional dimension, discomfort is entirely centered on the above-mentioned sense of deep solitude and seclusion lived in prison and the difficult condition that prison life imposes. The feeling of loneliness [20,21] dominates the narratives together with the widespread fear regarding health [46].
Illness, in fact, is present in prison even when there are no severe pathologies, but rather psychological discomforts like the "fear of getting ill". Environmental, material (humidity, forced cohabitation, hygiene) and psychological conditions (physical constriction, lack of affections) influence the individual experience and, therefore, psychological support becomes more urgent than material support. According to Cowen [13], environment is a basic issue to understand the level of empowerment experienced by the individuals.
Here below are the answers to "Do the living conditions here (environmental, hygienic etc.) influence your health and well-being?": "Yes, the conditions of life here influence our state of health greatly, especially at the psychic level, there is a lot of stress here".
"I am convinced that prison, living here, greatly influences our health, regarding me, very much, because as soon as I came here I had depression problems, many of us suffer from depression here. It was hard, actually, it is hard".
"I am convinced of this, according to me, the conditions of life here influence our health, look at me, I was in good health out of here, I got ill here".
"Yes, there are some inmates that have quite serious illnesses even if I am healthy and I am counting on leaving healthy from here. There was an inmate here with Aids before, she was honest about it and told us right away, many have hepatitis".
Cohabitation in the same cell with people with transmitted diseases, the general hygienic and environmental conditions of prison worry the detainees who believe they do not have sufficient information regarding eventual prevention measures. However, all the detainees refer to an "information sheet" which is given to them when entering prison and in some (rare) cases a course on prevention measures is organized ad hoc. Nevertheless, it is mainly their own common sense and self control which help the prisoners to avoid the risk of being contaminated.
Here are the answers to "When you entered prison, were you given general or specific information on disease prevention?": "I must say no, there is not enough information, when you come in here, they do not tell you to be careful here and there, everyone here has to take care of themselves because we all know we must be careful, now they are doing prevention programs for cancer, Aids, hepatitis and so we are better informed".
"No [no information at the entrance] not even here, no. In fact, the questionnaire I filled in, I answered in any case. It would be nice if we had a meeting with the nurses too. Well, with the people with hepatitis, I do not drink in the same glass, this is normal, but I… that is, I am a mother, a grandmother, I have experience and know these things, but what happens when a young girl of 18 comes here and has to take care of herself ?".
Not having the possibility to choose their own doctor nor the time for their treatment, they find themselves forced to face the difficult moments and have to trust the prison staff available totally.
In this sense, the trust that doctors are able to establish with the prisoners seems to be decisive in regards to the perception that the treatment prescribed is effective or inadequate. Likewise, communication becomes a crucial point to establish the doctor-patient relations, as asserted in a more general sense by Zani & Cicognani [33] relatively to "emotional" and "informative" forms of support. In fact, inadequate treatment is often associated with insufficient information in our narratives. Therefore, the prisoners who are able to understand that, for example, the waiting period for a medical visit is due to bureaucratic reasons and not to the disinterest of the medical staff, live their condition better. Besides, a very urgent request driven by prisoners is that related to an increased presence of doctors, psychologists and nurses.
Here are the answers to "What is your relationship with your healthcare staff like? Do you trust them? Why?": "I have a good relationship with the doctors and nurses, yes, I trust them, but I have to, it is not like out of here where you have your own family doctor, so, I must say I trust them".
"I needed a recovery, I also applied for examinations by a consultant, but nothing, they refused. I have blood tests but they never let you know anything, they could at least let you know the results, instead nothing".
"Not only for me, what I am asking for is more visits for women, especially gynaecological visits for the women who need it, on a regular basis not only when you ask for it, visits that women should have to check their uterus, their breasts. Because what was not there yesterday can be there tomorrow… and then when we ask for a medical visit they ask us what problem we have and according to what you say they give you a pill, and then that pill could be bad for you. I cannot say "I have a cough" and they give me an antibiotic, because I might not have bronchitis but something else. The health care officers should be more present".
In general, the level of treatment received is considered good, but -as said suprathe evaluation is strongly influenced by the affective and personal relationships between the patient and medical staff. The internal treatments are principally considered good, not too different from those received outside. The problem arises when there is the need for a specialist visit, considering the waiting time (code: "too long waiting time"), for the double bureaucratic problem regarding pharmaceutical stocks and availability of specialized interventions.
Here are the responses to "Do you think these treatments are well-timed/fast/ efficient?": "I believe that the treatments are the same as out of here, because I am continuing the same treatment, I am followed by them, but I am sure that outside things are different, because you are outside, you are not forced to live here, and it is the environment that changes everything here".
The correlation between the dimensions connected with health (which we coded as "cure") and the affective dimension upholds the consideration that moral support from the medical staff and communication are two important factors for the prisoners' well-being. Discomforts linked to prison life become extreme because of the condition of seclusion, which induces affective vulnerability. According to these findings, in the theoretical framework we saw that in general relatedness strongly influences happiness, according to Argyle [14]. Dialogue and empathy are fundamental indicators of relatedness, according to Reis et al. [15]; and specifically for the prison settings, we considered that Toch [29] recognizes "support" (e.g. responses to needs for understanding, empathy, warmth and emotional support) as one fundamental need for the prisoners.

Future objectives and the fundamental role of motherhood
It is interesting to observe that, although discomfort and loneliness are the most significant constant in the contents analyzed, with fear of illnesses as a recurrent theme, most of the prisoners interviewed consider their future quite positive.
The elements that determine optimism are the following: 1. family support: in many cases this regards women who are not very young and invest their future entirely in the family; 2. having participated in projects, training or educational courses providing skills to utilize outside; 3. motherhood, above all.
Motherhood proves to have a direct influence on the motivations towards the future. It seems to be an excellent sign in the task of serving time in prison and continuing eventual treatments after detention. The sense of responsibility towards their children becomes functional to the assumption of responsibilities towards themselves.
The element of external support and, above all, the presence of children is one of the women prisoners' motivations to resilience, as seen by the relations between the codes "Motherhood" and "Family" and the projection of the future (positive/negative). Unifying the two codes through the AND Boolean operator, a dimension of great confidence in the future emerges, considering that all the quotations concern the code "positive future".
Here we can see the answers to "How do you see your future?": "My future looks positive, I pluck up courage by myself to carry on every day. This experience has marked me, it has made me understand many things, and I often talk to my children about it, I don't want them to make my same mistakes, they must leave here, from Naples. They are young and can still save themselves, I have paid for all my mistakes, they must not live the life I did, I want them to leave, they must start a new life".
"I imagine my future positive, beautiful, at home, with my husband and children, I imagine it positive, especially for them, I want to make up for the time lost, I want them to have everything, I want to be present every moment, because this experience has made me realize the importance of life, the sense of a family, now I appreciate all that I had before, and when I was out I did not even consider".
Family or motherhood are never considered a problem when referring to the period of regained liberty.
Studying this aspect in depth, through crossbreeding with the Boolean operators, we notice that in the interviews a more frequent relation between the idea of a positive future and the presence of a support outside emerges more often when children are present (example of a quotation: "my future are my children"): social rehabilitation is considered possible and actual in this condition.
The most frequent quotations recall, on one hand, the will to offer their children more and better possibilities ("think of my children, save them from this world"; "my children have to leave from here"). On the other hand, the idea to be taken care of by them, somehow protected from the possible difficulties of post-prison integration ("I will think about my family and nothing else"; "I have a family that is waiting and supporting me outside, therefore I feel at peace"; "I see my future as a grandmother").
In the presence of the family as an external support, the interviewees declare to imagine their future as totally centred around their family life. This offers them a cognitive tool to invest their energies positively and not let themselves go into depression or discouragement for their imprisonment.
Considering the condition forcing them to be far from home and the bad example given to their children ("you are here and you feel a shitty mother"), they do not want their children to have their same prison experience: "they do not have to live my same life; seven children, all with clean records; It is terrible to have to tell your little girl that her mother is in prison. I have not seen her for 1 year, because I do not want or however I am still not ready to receive her in this infernal place".
In fact, two types of problems are considered by mother prisoners: 1. firstly, the difficulties regarding family visits and the possibility of allowing their children to live moments with all the family united. 2. secondly, the absence or shortage of activities for them as a family, which prevents the realization of Toch's need for activity [Ibidem].
In the first case, for example, the interviewees refer to the possibility of spending time in the green area during visits, but also to the prohibition of the father participating in the (monthly) scheduled meeting hour between mother and child: "We have to recreate a home environment in a green area, so why can't my children's father stay? Are we a family or not?".
Moreover, the absence of ad hoc courses or facilities to share the growing up period with their children causes a deep sense of failure of the maternal role. This leads women to think about their children nostalgically most of the time, thus exasperating their condition of discomfort in prison life. This lack is caused by contingent causes, but above all by structural factors, i.e. the political choices in management of budgetary resources as Elaine Lord asserted for U.S.A. mentally ill prisoners (but the situation is very similar in Italy): "given that more than 90 % of the budget of any facility is expended in security, this left very little funding for programs to address the underlying issues of the mentally ill" [(47): 940].
Here are the answers to the question "Are you taking part/did you take part in intramural activities?": "How do I feel? I feel… all right this is my status quo, anyhow, I feel like I do not care anything about myself and here…[crying], the issue mother/daughter always comes out and I said, I feel like shit…as a mother, so…maybe many people live motherhood differently than me. My strength inside here, I get it from them because inside here there is nothing that gives me strength, so if I feel bad I keep it to myself". "Here, I do not spend my day in the best way, because I always have my mind on somewhere else, at home, on my children, but to take my mind off this for a while, to escape from my thoughts that accumulate, I attend courses to help me relax".
Obviously, a mother's imprisonment directly influences her child's life, which becomes more difficult when very small children have to experience imprisonment with the mother. In fact, it seems that appropriate facilities and proper supports for this cohabitation do not exist. In a case narrated by a mother prisoner with a baby of a few months, she said that the treatment for the child was absolutely insufficient and inadequate to monitor his physical and psychological growth during the first months of life, and that the availability of activities or a proper environment for this circumstance did not exist, even when foreseen by the Institute.
Here we can see the replies to "Tell us about a typical day here in prison": "For my child I do not get the cure he needs".
"There is no support among the women prisoners. Only groups are formed, therefore you must not talk to that prisoner otherwise you will risk being hit by that other one. You are threatened, but I do not care, therefore, if I feel like talking to someone I do it unhesitatingly. The only thing that I feel sorry for is the baby who suffers prison life conditions. I frequently ask myself what he is thinking, how he is living this period inside prison. I always try to transmit peace and tranquillity, even if inside here it is difficult. For example last Wednesday, a prisoner who looked possessed, for the love of another convict at the other building wing (where they await trial) started to scream like crazy, she started banging everything, four officers had to intervene to stop her and after hitting her, put her in solitary confinement. While all this was happening I tried to make my child think it was all a game, but it was very hard. Then, what makes me feel really bad is the moment we have to be locked in the cell, he starts crying. It is not easy at all, I hope they will let me leave. They could make me wear an electronic tag, force me to stay inside, move only from the bathroom to my room, but the important thing is that I get out from here".
International research confirms the central role of motherhood in prisons [11], showing that mothers consider separation from their children to be the hardest aspect of imprisonment [48]. They often feel "enormous grief" about the time lost with their children. Children continue to have a fundamental role in women's lives during incarceration, especially because they are considered the sole and true motivation for change and their primary purpose in life [42].

The principal correlations emerged
Related to the theoretical framework described and linked to the scientific literature in this research field, the principal correlations which emerged from our research confirm a few findings briefly reported below: Positive Emotional Dimension is strongly correlated with: 1. External support, especially in relation with family support, according to Argyle [14] and Enos [11]. 2. Internal activities, particularly courses and projects, according to Carcedo et al. [25][26][27]. These activities are fundamental to fight loneliness as social isolation, according to Weiss [21] and Toch [29].
Positive Future is strongly correlated to: 1. Self-efficacy, active when the daily activities are meaningful and goals are realistically attainable, according to Brunstein [17] and Dodge & Pogrebin [48]. 2. Motherhood, considered as sense of responsibility, according to Fogel [39] and Ferraro & Moe [42], and also to overcome the feelings of guilt for the failure of the maternal role and time lost with their children, according to Fogel & Martin [38] and Lindquist & Lindquist [37].
Health and Well-being, in an "eudaimonic" sense [12], are strongly correlated to: 1. General prison environment, linked to the nature of prison regime, according to Cowen [13] and Biggam & Power [35]. 2. Relations with officers and health professionals, according to Reis et al. [15]. 3. Effective information at entrance on the procedures and health risks, linked to the presence of physicians and psychologists, according to Malloch [9].
These findings can be confirmed by the following networks, built through the Boolean operators (Figs. 1, 2, 3 and 4): Lack of external support is linked to: stigmatization/exclusion/prejudices; social disadvantage; being a woman in prison; anxiety; depression.
Fatalism is linked to: uncertain future; negative future; depression; long empty days. Internal support is linked to: relations with the officers and staff; relations with the health professionals; relations with the other prisoners. Loneliness is linked to: lack of psychological support; depression; long empty days; separation from the family.

Discussion and conclusion
Considering the theoretical model of the Self-determination Theory [12], data from our interviews show that social rehabilitation is experienced more peacefully when the family offers its support, thus becoming the driver motivating prisoners to face their detention period. As Barbara Bloom and her co-authors asserted: "the dominant theme of connections and relationship threads throughout the lives of women offenders […]. People in prison often live what Zygmunt Bauman [50] defines Unsicherheit, that is, the sum of their experiences defined by the words uncertainty, existential insecurity and unsafeness. The spiral triggered by these afflictions is a significant impediment to collective remedies. People who feel insecure, mistrust what the future holds for them, fear for their personal safety and seem paralysed by individual tasks to the point of not being able to imagine different ways of collectively tackling their problems.
This mental state of "anomic" alienation in relation to the experience of time in prison is well explained by Erving Goffman, who, in Asylums [30] wrote how inmates feel that the time spent in prison is wasted and useless. It is a period of time that seems to be "hibernated", something that must be "passed" or "marked" or "delayed"; it is basically a period, which elongates and retracts itself without passing, and that fundamentally marks the time imposed by the authorities. In fact, "under the prison system, dependence on authority figures is maximized, and opportunities to learn and experience responsible personal decision making are minimized" [(51): 46].
The most significant findings which emerge from data offered by the interviews can be identified in the core concept of loneliness, which is typical of seclusion. It is the greatest difficulty of penitentiary experience, which can degenerate in the absence of a stable affective network or of trust in the prison staff. As we saw in the literature review, emotional, instrumental and informative support [33] are fundamental to cope with the difficulties in the total institution.
Motherhood, in particular, becomes the most delicate issue for women in prison. The maternal role is lived as frustrating due to distances and the actual obstacle in maintaining their educational role in their children's lives. This is the reason why all mother prisoners think about their future as a redemption of their motherhood and intend to spend their liberty period dedicating themselves completely to their family.
Children at home may scarcely influence mother-prisoners' mental health: although regular visits can maintain family links, the frustration and worry of being unable to interact freely with their children may cause great distress and sometimes desperation for the prisoner. This is often exacerbated by the distance from home due to the comparatively small number of women's prisons, which often results in considerable travelling for their families. The distress experienced by these mothers is another factor raising the risk of self-harm [(10): 165].
Our interviews show multiple levels of reflection, crossbreeding mainly two plans: a situational and a personal plan. The main dimensions emerged are: access to activities in prison (work, projects, sociality) experienced as a deterrent to seclusion and as an opportunity for future reintegration into the "free" society. Motherhood is a spur to resilience, but also in reference to the difficulties associated with the conditions of segregation. The main theme that crossed all the conversations is "loneliness", lived as a "prison disease".
From an interesting survey carried out on female offenders in U.S.A., it emerges that seven strategies are actuated by women prisoners respect to their motherhood: being a good mother, disassociation from prisoner identity, mothering from prison, role redefinition, self-transformation, planning and preparation, and self-blame [52].
Being a good mother is a coping strategy activated by the women to affirm their fitness as mothers.
Dissociation from prisoner identity consists in minimizing "prisonized" behaviours, thereby distinguishing themselves from other prisoners and from a prisoner's image.
Mothering from prison means that they utilize this strategy to sustain their maternal bond maintaining contacts with their children and caregivers.
Role redefinition is a more complex strategy, denoting that children take the roles of parents or peers. So, "by attributing exaggerated maturity to their children, these mothers seemed to neutralize the harms they may have caused their children, while minimizing their own guilt and sense of failure" [Ivi: 462].
Self-transformation, which consists in processes such as becoming spiritual or religious, or being involved in community and helping others, in a sort of "conversion".
Planning and preparation for the future (getting jobs, finding places to live), as adaptive strategy for surviving.
Self-blame, that is admitting feelings of guilt and shame and expressing responsibility for their past behaviour. This strategy may be maladaptive or adaptive; therefore, it "might be self-harmful, especially when no help or assistance is provided to mothers to 'counterbalance' it. On the other hand, self-blame, if followed by real opportunities for self-transformation and a change in circumstances, might be a step toward positive reinterpretation and adaptive coping" [Ivi: 465].
The prisoners of our survey seem to principally adopt the last two above described strategies. They live their life manifesting many fears regarding their well-being in prison, above all, in regards to the upbringing of their children, having to cope with feelings of guiltiness because they have deprived their children of an important source of material and emotional support.
The number of prisoners in the institutions, the continuous-change over due to short sentences and to concessions of alternative measures and the different social-cultural conditions which distinguish them, make the programming of valid rehabilitation and reintegration activities more and more difficult. In fact, considering the intuitions of the Canadian sociologist Erving Goffman [30], total institutions are incompatible with a fundamental element of our society: the family. Therefore, more attention should be given to the re-socialization aspect of prison. Furthermore, problematic aspects of prison life, such as the distance of women's prisons from cities, costly phone calls, rigid rules of visitation, and so on should be reconsidered [(52): 468].
It is clear that in order to interrupt the vicious circle of social exclusion -prisonnew social exclusion, complex interventions supporting gender perspectives are necessary. Therefore, consideration must be given not only to the profile of women prisoners, but also to their backgrounds of social discomfort and exclusion conditions before imprisonment. The creation of support networks, integrated between public institutions and Non-Governmental organisations become a central element and, somehow, predictive of a possible and positive reintegration into society. Restarting from prison means ensuring respect for all fundamental rights outside and inside prisons, promoting social policies and reducing situations of discomfort in order to eliminate the causes "creating" imprisonment.

Examining trafficking statistics regarding Brazilian victims in Spain and Portugal
Introduction A debate about human trafficking almost always involves data which showcases its alleged gigantic proportions. The existence of such numbers is not in itself a negative thing, as statistics about a criminal phenomenon are crucial to ensure that the policies to combat it are both proportionate and effective. The vast majority of existing trafficking data, however, does little to clarify the reality of the situation. The fact that trafficking victims mostly consist of a hidden population is used as an excuse to project the existence of an immense dark figure which dwarfs the number of detected cases. The figures which are inferred, often from a handful of confirmed victims, are primarily used not as an honest assessment of the issue, but as a way to stir indignation against it. This is very clear when we examine statistics which refer to Bpotential victims of trafficking,^addressing not only people who show signs of being in a trafficking situation, but also entire subgroups which are labelled as being Bat risk^due to their supposed vulnerability. Considering that these groups often consist of sex workers, irregular migrants and minorities, as well as women in diminished socio-economic circumstances, it is no wonder that the alleged figures being bandied about are so large. 1 Very little, if anything, is said about the methods used to collect the numbers which are presented. Moreover, it is often unclear what exactly is being counted as a (potential) victim of trafficking, which may vary from those who fulfil the standards of the United Nations (UN) Trafficking Protocol 2 to all kinds of forced labourers, irregular migrants and migrants who sell sex.
The lack of scientific scrutiny regarding most of these numbers is disconcerting. Statistics are often repeated ad nauseam without any reference to their sources or the years they were developed. They are often misattributed to large international organisations, such as the UN, as this apparently gives them legitimacy. Significant weight is given to unsubstantiated media reports. This trend, Bwhere tabloid journalism is footnoted, referenced, and hence legitimated^( [42]: 417) is rightly vehemently and universally denounced in other areas of academia, but is often accepted without question when human trafficking is being debated.
In the past few decades Brazil has gained and maintained an international reputation as a major source country of trafficking victims, particularly to Europe. 3 Yet despite the continuous avowal that transnational trafficking is a large-scale problem in the country and increasing efforts to combat it, statistics about the phenomenon are scarce and questionable.
The lack of valid data has not stopped the dissemination of a variety of assertions about Brazilian trafficking victims. One of the most common is the insistency that there are large numbers of (female) Brazilians trafficked to the Iberian countries, which are supposedly among the main destinations for Brazilian victims. 4 Even though this premise has been the basis of a number of bilateral debates, agreements and joint operations between the three countries [68], there are few data sources regarding Brazilian victims in the Iberian countries that can undergo any sort of scrutiny and the numbers produced by those tend to be rather small for a phenomenon which is supposed to possess epidemic proportions.
Although an extensive literature exists on the myriad problems which affect trafficking data in general [57], less work has been done on dissecting the existing problems of data in specific countries. This avenue of exploration is important, as challenging the numbers which are used to shape a country's trafficking policies may also mean challenging the policies themselves. Thus, this article will address some of these data issues in relation to Brazilian victims of transnational trafficking, focusing particularly on victims of trafficking for commercial sexual exploitation that end up in Spain and Portugal. 5 The aim of the article is not to refine existing data to obtain as accurate a report as possible regarding the number of Brazilians victimised by transnational trafficking, but to work with the existing publically available data and analyse its validity.
A brief introduction will, based on existing literature on the subject, highlight the importance of (comparable) trafficking data, the most common problems with data collection and proposed solutions to improve the data gathering processes. The focus will then shift to two particular issues: the internal inconsistencies of the transnational trafficking data in Brazil, Spain and Portugal and the cross-comparative inconsistencies of the data regarding Brazilian victims of trafficking between these countries. The methodological issues present in some of the data gathering processes and their consequences to the understanding of human trafficking in this scenario will be addressed. The article will then conclude with the presentation of possible obstacles to establishing strong and reliable data gathering systems.
These data assessments will be based primarily on an extensive qualitative data collection undertaken by the author which took place between January of 2012 and January of 2014. This data collection privileged primary sources produced by relevant government organs (police forces, migration offices, courts, etc.) and international organizations (primarily the UN and the European Union) which are available to the public. 6 When primary data was not accessible, efforts were made to find reliable secondary sources of information that not only clearly established that the data was obtained from governmental institutions, but also expanded upon the methodology which was used when collecting it.

The need for accurate statistics
Accurate statistics do more than give a more precise dimension of the extent of human trafficking. They enable both governmental and non-governmental organizations to adequately tailor their responses to the phenomenon regarding, for instance, needed financial resources, as well as suitable facilities and services that need to be available to victims. They also allow governments to produce evidence-based trafficking policies and assess throughout time whether these policies have been effective [57].
Bad statistics, therefore, may do more than impede an appropriate allocation of resources; they may also shape and support bad policies. In fact, such statistics may actually be deliberately constructed to stir fear and indignation and through this enable 5 Primary focus will be given to adult victims of trafficking, as the criteria established in the UN Protocol for labelling minors as victims of trafficking is different. Nevertheless, in some cases the presented data conflates adults and minors, without taking into consideration the differences between them. This amalgamation, if it takes place, will be explicitly stated. 6 The focus of this data collection is on statistics produced and/or sponsored by national governments and/or international organizations, as this is the data which shapes and is shaped by official policies. This means that data collected by non-governmental organizations (NGOs) will not be taken into account unless it is explicitly government endorsed. the shaping of trafficking policies whose primary concern is not victim protection, but the constraint of migration and sex work. 7 Having accurate and internally consistent data be available for individual countries, however, is not enough. As trafficking is a crime with extensive transnational dimensions, it becomes imperative for the data produced to be comparable in a crosscountry basis. This compatibility, however, is notably absent in the vast majority of cases.
One of the primary reasons behind data incompatibility is that although a lot of statistics claim to be assessing Bhuman trafficking,^not all of them are identifying trafficking victims per the full understanding of the UN Trafficking Protocol. In some countries the distinction between trafficking and smuggling may be blurred. In others, no distinction is made between victims of trafficking and migrant sex workers and other forms of trafficking besides sexual exploitation are completely ignored. In many more, the concept of (abuse of a) Bposition of vulnerability^is interpreted in such a way that poverty is seen as a form of force and therefore the non-autonomous migration of poorer people (particularly women who migrate and perform sex work) is interpreted as being a form of Bstructural trafficking^ [50].
Some of the time, such interpretative problems are a result of unclear or outdated legislation, making a shift towards the use of an UN-compliant definition of human trafficking a paramount priority. 8 Other times, however, such conflations appear to be intentional. Countries that have an abolitionist or prohibitionist position, for instance, deliberately make no distinctions between all migrants who sell sex, as it enables them to use trafficking policies to target migrant sex work. In such cases, there is little internal motivation to improve the legislation.
A second crucial step is the need for a centralized data gathering system which will coordinate data received from several distinct sources into one single product (be it a database or periodic report). This will not only ensure the production of internally coherent data, but will also make access to such data much easier and, if the appropriate care is taken, minimise, if not entirely eliminate, the possibility of double counts.
To be of utmost use, trafficking data should also be publically available. There is, of course, the need for some of the information to remain confidential due to privacy and security considerations. Nonetheless, most findings (such as overall numbers of victims, as well as types of trafficking and nationalities detected) can be released without greater problems. Considering the invaluable role such information may play in the actions of governmental agencies, the development of policies in other countries and in the decisions undertaken by auxiliary support systems (which include everything from local NGOs to international organisations), it is not unreasonable to assume that the time and expenses incurred in the production of an annual report or similar data compilation are more than worth it.
Presenting such information publically, however, is not enough. Transparency is also of the utmost importance when it comes to explaining how the data was obtained. Who was responsible for gathering data, which criteria were used to separate valid from invalid cases and, most importantly, which definition of human trafficking is being used are some of the key pieces of information that need to be made available so that the validity of the presented statistics can be properly considered.
Although these changes may seem daunting, significant efforts have already been made to establish guidelines for integrated trafficking data collection systems ([1]: 93-106) by both international organizations and individual states. Some of these efforts are particularly relevant to the current discussion.
The Following a similar pattern, in 2009 the International Organization for Migration (IOM) produced a set of BGuidelines for the collection of data on trafficking in human beings, including comparable indicators.^It provided templates for more comprehensive data collection on victims, traffickers, the trafficking process and criminal justice responses to trafficking. [1] Its results, however, were similar to ICMPD's.
Under the aegis of the European Commission, the Institute for International Research on Criminal Policy (IRCP) developed in 2006 SIAMSECT, a set of standardised templates for a EU-wide collection of statistical information regarding missing and sexually exploited children as well as trafficking in human beings. Following an event, victim and author template, SIAMSECT allowed multiple ground actors (not only law enforcement and judicial authorities, but also other governmental offices and NGOs) to provide information which would go on to be centralized by a National Focal Point and then later on by a European Monitoring Centre. In this way SIAMSECT established a guideline for the creation of integrated and comparable statistical information at both the Member State and EU level, allowing for more effective cooperation within the EU, as well as between its bodies and other organizations. The SIAMSECT templates became the basis of the IRCP's 2010 elaboration of the MONTRASEC (Monitoring Trafficking in Human Beings and Sexual Exploitation of Children) demo, an interface and reporting tool which enables users to create an integrated data collection that gives an accurate picture of the relevant phenomena while following strict data protection guidelines [75].
The publication of Eurostat's trafficking statistics [13,14] showed, however, that although the ICMPD, IOM and MONTRASEC models were taken into account in the development of the EU's official trafficking statistics, none of them have been fully adopted in the EU's trafficking data collection efforts. Thus, even though the EU has recognized the need for comparable trafficking data since 1997 [75], it is clear that by 2014 problematic statistics were still being produced and disseminated in the EU and beyond.

Internal inconsistencies in Brazilian trafficking data
As previously stated, one of the major challenges of dealing with human trafficking is the establishment of a definition of the phenomenon. Although Brazil ratified the UN Trafficking Protocol in 2004, its current definition of trafficking is still not UNcompliant, as it does not address organ removal, forced labour or any other forms of trafficking besides those involving sexual exploitation.
Even the existing definition of trafficking for sexual exploitation is problematic. Prior to 2005, trafficking as laid out in article 231 of the Brazilian Penal Code (BPC) consisted of any sort of aid to the movement, to or from the country, of a woman who would end up working in prostitution. No specific means such as force or fraud were needed, nor was any kind of exploitation a prerequisite. In 2005 the legislation was modified for the first time and its two major changes included the expansion of the crime from trafficking in women to trafficking in persons and the introduction of Article 231-A, typifying internal trafficking. In 2009 the term Bother forms of sexual exploitation^was added to the law, which, however, continues to condemn not Bthe exploitation of the prostitution of others,^but prostitution in itself. As it stands, Brazilian legislation considers any kind of migrant prostitution which is not completely autonomous as human trafficking.
The first attempt to construct a country-wide diagnosis of the situation of human trafficking in Brazil was the BStudy of Trafficking in Women, Children and Adolescents for Commercial Sexual Exploitation -National Report -Brazil,^better known as PESTRAF, which was commissioned by the Organization of American States. Despite claiming to represent a Bstrategic investigation for developing new political practices that support the process used to resolve this problem^( [35]: 23), there seems to have been no concerted effort in the research to develop accurate statistics regarding the number of actual trafficking cases which took place in Brazil or involved Brazilian citizens abroad. The study did, however, attempt to map the existing international routes being used to traffic Brazilians (Fig. 1). In this mapping Spain was very clearly the country most paths lead to (32 routes), with Portugal trailing behind in fifth place (8 routes).
Source: PESTRAF ( [35]: 81) Fig. 1 Main trafficking routes of Brazilian victims by destination country. N/A represents the 33 out of 131 international routes Bmapped^by PESTRAF which were not given a specific destination country Although there is no supporting evidence regarding the number of persons allegedly trafficked through each route, the study strongly implies that there is a correlation between the number of routes and the number of victims. Thus, in PESTRAF's vision the higher the number of routes towards a country, the higher the number of victims in it. The possibility that a single route may be used by more people than multiple other routes seems to not have even been considered.
Even worse than the unsupported assumption that more routes mean more victims are the severe methodological failings which permeate the study [5]. When it comes to the alleged number of trafficking routes, this becomes particularly problematic on three accounts. The first issue is that the definition of trafficking used by PESTRAF considers both consent and coercive means to be irrelevant. Thus the routes which are laid out may have been used not by trafficking victims, but by completely independent migrant sex workers (a category whose existence PESTRAF, which classifies all prostitution as exploitative, chose to ignore). Secondly, the Bmapping^of the routes consisted in a mere compilation of information presented in unsubstantiated media reports, which often fail to make basic distinctions between trafficking in persons and migrant smuggling. Finally, one of the study's coordinators has admitted that their primary concern was not to provide data as accurately as possibly, but to stir society to react against human trafficking ([52]: 370), which puts the findings produced by the report, particularly those which claim trafficking is a problem of major proportions, in a distinctively dubious light.
Despite its numerous failings, the Bfindings^and assumptions from PESTRAF were and continue to be used as a source of information and policy guidance in Brazil [37]. Throughout the years information from the study came to be attributed to the UN (particularly the United Nations Office on Drugs and Crime -UNODCwhich adopted it as fact), further legitimizing it [5].
A couple of years after PESTRAF was released, a second set of data begun to circulate in a booklet produced in partnership by the Brazilian Ministry of Justice and UNODC [39]. One of its key pieces of information consisted of Bthe main destination countries^for Brazilian trafficking victims (Fig. 2). The booklet offered the percentage of victims which ended up in each destination, but did not clarify the total number of victims the percentages represented, nor how or when the data was obtained.
According to Blanchette and da Silva ( . These cases were not always confirmed as trafficking by the Brazilian judicial system and even when they were did not always fit the definition of trafficking established by the UN Protocol. The group of alleged victims also included a small number of minors, who were for the most part bundled with the adults. Just as problematic as the presentation of highly questionable and limited statistics as a Bcountry-wide diagnosis^of the trafficking situation are the inferences made from such data. The booklet claimed, for instance, that the Iberian countries (which it labelled as BLatin countries^) were the main destination for Brazilian victims of trafficking due to the language similarities. In its source report ( [8]: 37) the author went further and claimed that victims were Blimited^to Bneo-latin^countries due to their lack of formal education, which he imagined would preclude them from moving to countries with substantial linguistic and cultural differences.
There is nothing in the study to show that this was more than a mere assumption that ignores a number of issues. The first is that linguistic and cultural isolation are recognised as effective forms of controlling trafficking victims ( [56]: 107), which would seem to make traffickers' overwhelming preference for countries where this isolation will be somewhat limited suspect at best. Secondly, if victims were in fact Blimited^to countries with similar linguistic and cultural backgrounds, one would then expect to see an influx of victims to Latin American countries, which are notably absent from the list presented. The counter-argument that these countries are simply not rich enough to attract victims would not hold weight, as it would ignore the reality of South-South migration flows and the acknowledged existence of trafficking routes from Brazil to its (poorer) neighbouring countries. 9 Finally, this premise also ignores the massive flows of Brazilian migrants towards places such as the United States and Japan, countries which present significant linguistic and cultural barriers.
In the years following the presentation of this Bdiagnosis^Brazil continued to give increased attention to human trafficking, producing its National Trafficking Policy in 2006 and its First Trafficking Plan in 2008. Both were developed under the assumption that transnational trafficking was a major issue in Brazil despite the continued lack of valid statistics to confirm this assertion.
After the establishment of the First Trafficking Plan, two sets of data related to trafficking in Brazil surfaced in UN trafficking reports. Although not explicitly government-endorsed, the UN's trafficking reports and data have been considered to be reliable by those involved in the development of Brazil's trafficking policies [5]. The 9 See for instance [33] for a detailed report of the trafficking of Brazilians to Suriname.  [70] 2009 Global Report [70] presented some very specific trafficking numbers for the 2004-2007 period (Table 1). There are, however, major issues to be considered regarding the presented data. The first is that the 2009 UN report did not clarify whether it was addressing solely transnational trafficking or whether victims of internal trafficking (a crime that was typified in Brazilian legislation in 2005) were also being taken into account.
A second problem is that although the Brazilian trafficking legislation changed its definition of victims from women to persons in 2005, this dataset continued to show a clear focus on women and girls. Whether this was due to a choice by the report writers or the fact that the courts have only dealt with female victims (which is possible considering that the only type of exploitation recognised is that which is sexual and tends to be more female-centric) is not clear. The conflation of Bwomen and girls,â lthough a reflection of the Brazilian legislation (which establishes a unified threshold of victimhood), is also problematic, as the UN Protocol established different standards for the trafficking of adults and minors.
Another issue is that the quoted source of BFederal and State Courts in Brazil^is in practice very vague, as it does not clarify from which courts the data comes from or if all relevant courts have been taken into account. Considering that there is no centralised data gathering system for the Brazilian judiciary, there is a possibility that some cases are not being included.
Finally, while the UN is considered an important partner and source of trafficking information by the Brazilian government, other pieces of data presented in the report (specifically the number of cases of human trafficking investigated by the Federal Police 10 - Table 2 and those who have reached the courts - Table 3) do not match the data presented as official by the Brazilian government itself (Tables 9 and 10 respectively), despite the fact that the Federal Police is quoted as the source in all cases. 11 The problems did not end with the release of the subsequent UN report. In its BCountry Profiles -Americas^annex, the 2012 Global Report also presented some aggregated information about alleged cases (Table 4) which is quoted as coming from the Brazilian Public Ministry, the country's body of independent public prosecutors.
Although this information is offered in a trafficking report, it is important to keep in mind that while recognised in both its transnational (Article 231 of the BPC) and national (Article 231-A of the BPC) forms, the only type of trafficking which is typified 10 The Brazilian Federal Police is always responsible for investigating cases of transnational trafficking, while cases of internal trafficking are usually, but not always, assigned to the police forces of individual states. While the presented data most likely refers only to cases of transnational trafficking, we cannot affirm that it does, as internal cases involving multiple states may also be assigned to the Federal Police [37]. 11 It is important to bear in mind that the number of cases does not necessarily match the number of trafficked persons, as each case may involve multiple victims. Examining trafficking statistics as such in Brazil is that for the purpose of sexual exploitation. Thus, the cases of slavery and organ removal most likely represent the total number of such crimes detected in the country, which may or may not have involved trafficking. Moreover, it is unclear whether each Bcase^of trafficking corresponds to a single victim or if the cases refer to trafficking operations detected which may each contain multiple victims. As an additional problem, it is not clear what is meant by Bextra-judicial^cases of trafficking, although it may refer to cases which were investigated but did not produce enough evidence to go to court. The 2012 Global Report also presented additional data regarding Federal Police investigations (Table 5), which once more is inconsistent with official Brazilian data which was published afterwards (Table 9).
When it comes to human trafficking information in Brazil, investigative data seems to be the most easily obtainable. The oldest and most consistent figures available before 2013 referred solely to the number of Federal Police investigations related to Article 231 (Table 6). This data set has often been presented as official trafficking figures from Brazil and is the only piece of Bconcrete^(i.e., not based solely on estimates) trafficking data presented during the Parliamentary Inquiry Commission to investigate Human Trafficking which took place in the Brazilian Senate from 2011 to 2012 [63].
This data presents a number of problems. First, it refers to the number of investigations, which can and often do include multiple potential victims, who may not end up being confirmed as actual cases of human trafficking. Thus, while the numbers may show an overall increase in the efforts to detect trafficking cases, they do not help in estimating the number of victims detected. 12 Furthermore, as previously mentioned the definition of trafficking typified under article 231 has changed twice throughout the years, which means that the crimes being investigated in 2010 are not necessarily the same as those being investigated in the 1990s. Finally, despite its recurrent use, it is incompatible with data later presented by the government as official (Table 9), as well as with data presented by the UN's 2009 (Table 2) and 2012 (Table 5) reports. 12 Judge Rinaldo Aparecido Barros, considered a national expert on trafficking, claimed in a 2012 interview that the Federal Police has numbers that support the assertion that there are 75,000 Brazilian women who are victims of trafficking for sexual exploitation worldwide [66]. No other sources, however, including official data presented by the Federal Police, seem to acknowledge this number. It is likely a misrepresentation of an often repeated figure that claims that there are 75,000 Brazilian women working in the European sex industry. This data, often quoted without sources, is attributed by Téchio [67]   It was not until 2013, over a decade after the introduction of trafficking as a major public issue in Brazil, that an official compilation of trafficking data which included actual estimates about the numbers of victims was released by the Brazilian government. The BNational Report on Human Trafficking: consolidation of the data from 2005 to 2011^ [37] (later supplemented by a report on 2012 data [38]) provided a wealth of information regarding both national and transnational trafficking ( Table 7).
These numbers are particularly relevant to contextualize the known dimension of human trafficking if we consider that a mere 3 years before the publication of this report the Brazilian National Justice Secretariat claimed that on average 60,000 Brazilians were victims of transnational trafficking each year [54]. Yet despite the report's stated intention of offering a Bpicture^of the trafficking situation in Brazil, there are multiple problems with accepting the presented numbers of transnational victims as definitive.
First, the numbers presented in the 2005-2011 report are internally inconsistent. The total of 474 victims given in the country segregated data-set (Fig. 3) differs from the 475 victims presented in the data set which separates the types of trafficking. This is explained in the report by the fact that one victim was subjected to both labour and sexual exploitation and is thus counted twice. No explanation is given in the text, however, for the fact the types of exploitation data-set gives a different distribution of victims in the 2009-2010 biennial (88 and 218 respectively) from the one given in the country segregated data-set (93 and 213 respectively). 13 Secondly, although the Ministry has clarified that it (along with the embassies and consulates it administrates) adopts the trafficking definition set out in the UN Trafficking Protocol (Ministério das Relações Exteriores, personal communication, August 5, 2013), Brazilian legislation only recognizes those subjected to sexual exploitation (Table 8) as trafficking victims.
Finally, the data only encompasses victims who have come in contact with the Brazilian consulates and official representatives abroad. Yet not all victims contact the consulates, sometimes out of their own free will (being under theat times justifiablebelief that consulates will be unable or unwilling to help them), 14 sometimes due to outside forces (such as being barred from doing so by the police forces which have Brescued^them). 15 13 An inquiry sent to the Ministry of Foreign Relations (Ministério das Relações Exteriores, personal communication, August 5, 2013) clarified that the numbers were different because some cases which took place in 2009 were not accounted for until the following year. 14 This particular perception seems to be widespread among trans individuals and sex workers and has been confirmed by documented cases where help has been denied at certain consulates ( [4]: 94). 15 See for instance accounts in [60] about police forces deporting trafficking victims without giving them the opportunity to contact Brazilian authorities. Even in cases where trafficking victims are officially labelled as such by foreign governments and enjoy the rights that the category confers, no mechanism to ensure that the Brazilian government is informed when a victim of trafficking is identified abroad appears to exist. Moreover, the victims that do come in contact with the Brazilian consulates are further divided into three categories: a) those who go to the consulates looking for information or aid but are not identified as having been trafficked, b) those who are identified, but only want information or documents and thus are not registered as victims, and c) those who need help returning to Brazil or temporary shelter and are thus identified and registered. Only those who belong to the last category are classified as victims of transnational trafficking in the statistics compiled by the Foreign Ministry ([37]: 32). 16 Although most of these issues explain why trafficking cases may be under-reported, it is important to keep in mind that problems with the existing system may also lead to over-reporting. Gabriela Leite, one of the foremost activists for sex workers' rights in Brazil, affirmed, for instance, that Brazilian prostitutes will at times present themselves as being trafficked if they come in contact with the consular authorities abroad because they believe they are more likely to receive a better treatment if they position themselves as victims [53].
Even though the government recognizes the fact that the provided statistics may only represent a fraction of the existing trafficking victims, little seems to have been done to improve the situation. Despite the findings of the Guarulhos Airport studies [60,62], which showed that it is possible to identify concrete and potential situations of trafficking through a consistent supportive approach to Brazilians who are deported or non-admitted abroad and thus forced to return to Brazil, the support that does exist is not dependable. While the Bhumanised assistance offices for migrants^1 7 have multiplied, their work still remains inconsistent and systematic and valid data does not seem to be collected by this auxiliary support system (D. Figueiredo, personal communication, July 18, 2013). 18 Following the trend of previous data sets, the National Report also contained data regarding police investigations (Table 9) and court procedures (Table 10). While the 16 It is relevant to consider that to be statistically valid the classification of the trafficking victim depends on the victim's agreement to the use of the label. One can argue that this may result in under-reporting of trafficking cases, as some victims do not identify as such. It is very important, however, to recognize that this goes against the widespread trend of ignoring the agency and autonomy of migrants by labelling them as trafficking victims, often to their detriment, without considering whether they see themselves as such. 17 These outposts are present along selected migratory hubs (mostly in major airports and in the Amazon) and are supposed to, among other things, provide information and support to previously identified trafficking victims and identify through interviews previously unknown victims that have been denied entry into other countries or deported. 18 A similar issue appears to happen with non-governmental organisations (NGOs). Although they are often relied upon to support government efforts (see for instance, the II National Plan on confronting trafficking in persons [61]), there appears to be no mechanism which enables data to be gathered about trafficking victims supported or identified by NGOs. definition used by the Federal Police in these cases appears to be consistent with the 2005 version of Article 231, the actual number of victims cannot be estimated because, as previously mentioned, each case investigated may contain multiple victims or none at all. The data available regarding the number of actual criminal indictments in the courts is significantly smaller (Table 10). There are two reasons for that. The first, which is not limited to trafficking cases, is the fact that not all investigations lead to the discovery of actual crimes. Even when they do, sometimes there is simply not enough evidence to take the cases to court. The second is the fragmented and at times inexistent sources of judicial data regarding human trafficking in Brazil. The data from Table 10 was compiled by the National Justice Council (the administrative organ of the judicial system) and came mostly from Federal Regional Tribunals (although ten cases were reported by Justice Tribunals). As there is no central national database or means for the Council to obtain the data itself, it was dependent on the contribution of individual courts, not all of which answered the information requests that were sent. A similar issue was encountered by members of Congress during the investigations of the Parliamentary Inquiry Commission on human trafficking [63].
The problems we face are therefore three-fold. First, officially endorsed victimrelated Btrafficking^data in Brazil actually consists of three different categories: victims who needed the resources of the Foreign Ministry, cases investigated by the Federal Police and cases taken to court. Each of these statistical categories is obviously shaped by the primary area of interest of the entity collecting data.
Secondly, the indiscriminate and at times unclear way the data is presented gives a confused and at times unclear picture of the magnitude of the problem in Brazil. Extrapolating the existence of 75,000 victims of trafficking [66] from an average of little more than 41 cases investigated a year (per Table 6), for instance, is highly problematic.  [37,38] Finally, even the data in these self-contained categories do not match. For the year 2007, for instance, the 2009 UN report acknowledged the existence of 5 female victims, per the Brazilian courts [70], while the Brazilian Foreign Ministry claimed to have identified 38 victims of trafficking in that same year [37]. The difference, although problematic, is understandable, as what is being labelled as a Bvictim^is different in each case. The courts were only considering women and girls who fell under the provisions of article 231 of the BPC (which differs from the UN Protocol trafficking definition by, among other things, bundling adults and minors), while the Foreign Ministry used the UN's definition to identify all Brazilian victims of trafficking abroad which needed the help and resources of its consulates.
The situation regarding the number of cases investigated by the Federal Police is even worse. As the parameters being used are presumably the same (cases investigated under article 231 of the BPC) and the primary source quoted is the same (the Federal Police), one would expect the numbers to match, which is not the case. For the year 2007, the 2009 UN report asserted that the Federal Police investigated 109 trafficking cases [70], while the 2012 report changed this number to 114 [71]. Data presented in 2011 by the Federal Police itself claims a middle ground with 111 cases having been investigated [63]. This represents, however, a stark contrast to the numbers quoted in the government's 2013 trafficking report as coming from the National System of Criminal Information and the Federal Police itself: 26 cases of trafficking investigated [37].   [37,38] Internal inconsistencies in Spanish data regarding Brazilian trafficking victims Efforts to compile human trafficking data in Spain date back to the late 1990s. Despite this relatively early start (which preceded even the UN Protocol by a few years), the trafficking data collection in that country presents a number of significant problems. It is important to consider, for instance, that the Spanish legal definition of human trafficking has not been constant throughout the years. At first, data labelled Bhuman trafficking^focused on coercion into prostitution, which since 1999 explicitly included, but was not limited to, trafficking for sexual exploitation. In 2003 the perspective about trafficking widened and started to include all cases of irregular migration whose purpose was sex work, even if there was no exploitation involved. 19 Specific legislation which clearly differentiated human trafficking from migrant smuggling as per the terms of the UN Protocol did not exist until 2009. It was not until the following year, moreover, that the legislation was modified to acknowledge all forms of trafficking rather than just focus on trafficking for sexual exploitation [50].
A second major problem is the lack of a centralised data collection system. Data is gathered by individual police forces (the Guardia Civil 20 and the Cuerpo Nacional de Policía 21 ) and the Fiscal Ministry 22 and later also compiled by the Intelligence Centre Against Organised Crime (CICO), the Ministry of Interior and the Attorney General's Office (Fiscalía General del Estado). Data from these sources, however, is not always consistently publically available.
Perhaps worse than the multiple data sources is the fact that the data is at times inconsistent among the sources themselves, which is exacerbated by the multiple definitions of trafficking in persons being used, even by single governmental entities, regardless of the actual legislation which is in place. Although each of the sources presented does elaborate on 19 This, however, excluded cases of trafficking where the victim was legally present in the country. This is particularly problematic when we consider that official data from 2010 implies that in over 65 % of the cases in that year where trafficking indicators were found the potential victims were regularly residing in Spain ( [58]: 44). This large percentage, however, may be a reflection of the fact that many undocumented migrants are summarily expelled from the country when identified by the police and not given a chance to have a potential trafficking situation investigated. 20 The Spanish Civil Guard has jurisdiction over the majority of the Spanish territory (although that amounts to but approximately 40 % of the population), including border posts such as airports. The rest of the territory is overseen by the Cuerpo Nacional de Policía (National Police Corps, responsible for urban environments) and the Autonomous Police Forces of Catalonia and the Basque Country ( [25]: 3). 21 Unlike the Guardia Civil, which counts all sex workers it comes into contact with as potential victims, data from the CNP supposedly only counts those who have been submitted to a situation of trafficking for sexual exploitation ( [68]: 93). CNP data, however, is not publically available and attempts by the author to obtain it were unsuccessful. 22 The Fiscal Ministry (known in other countries, such as Brazil, as the Public Ministry) is the body of public prosecutors. the fact that a small percentage of the victims are minors, this information is never segregated by country of origin and does not impact the final Bvictim^count, which always incorporates adults and minors as if there were no distinctions between them. Data specifically mentioning Brazilian victims, although sparser than overall trafficking data, is also available. It suffers from the same problems found in all of Spain's trafficking figures, with inconsistency resulting primarily from incompatible definitions of what constitutes a Bvictim of trafficking^being the most easily observable (Table 11). Besides the large differences between the data sources, it is interesting to note that they also fail to follow similar patterns regarding the increase and decrease of the alleged number of trafficking victims across the years.
Data about Brazilian victims is publically available from three major sources: the Civil Guard, the Ministry of Interior and the Fiscal Ministry. Due to the fact that the Guardia Civil has consistently published a (mostly) publically-available annual trafficking report since 1999, 23 its trafficking data is often used to represent the totality of trafficking cases in Spain despite the fact that it is not the only police force in the country and thus only deals with a fraction of the total of victims. 24 The Guardia Civil focuses the vast majority of its efforts on detecting trafficking for sexual exploitation which occurs in highway clubs. 25 By taking into account only its information, trafficking cases which potentially occur in street prostitution and those which take place in apartments are often ignored. This is particularly problematic when we consider that these forms of prostitution tend to be more attractive to male and trans 26 sex workers, who have little to no participation in highway clubs and are thus mostly invisible in the Guardia's statistics ([6]: 37). 27 Perhaps worse than that is the fact that the Guardia Civil presents in its reports two sets of data. The first set (row I, containing by far the largest numbers in the table) refers to Bpotential victims^as per the understanding of the Spanish Civil Guard: …although technically the denomination of Bvictim^should only be applied to a person that reports being sexually exploited, for the effects of this report [we] will consider Bpotential victims^the people (mostly women) who engage in prostitution in these establishments, because [we] understand that they do so without any administrative support, due to finding themselves in a situation of necessity (although not all of them) and because, making an accusation or not, there is always the lingering doubt of whether they engage in this activity under the greater or smaller control of people and organisations dedicated to illegal activities ([32]: 13).
Note that these Bpotential victims^need not show any of the usual signs of a potential trafficking situation, such as violence, movement control or retention of documents; working as a prostitute is enough. Nor is there any indication whatsoever that these potential cases undergo a scrutiny longer than a brief conversation between a trained police officer and the alleged victim to determine whether trafficking is indeed taking place. There is also no system to ensure that the same woman will not be counted twice. Considering the widespread plaza system in Spain, in which women usually spend 21 days in a particular establishment before moving to another, 28 the possibility of sex workers being counted multiple times, in different cities and even communities, is not a small one. Despite these issues numbers obtained through this understanding that all Bvictims of prostitution^are (potential) trafficking victims are often used to represent official Spanish trafficking statistics, particularly when there is a need to reinforce the idea that trafficking is a major problem in Spain. 29 The second set of data from the Guardia (row II) contains much smaller numbers which refer only to women who have claimed to have been exploited and/or forced into prostitution. 30 Despite referring to these women as the Breal^victims in its 2009 report 28 The plaza system's 3 weeks length is supposedly tied to the length of women's menstrual cycles, allowing, in theory, sex workers to complete a non-stop work cycle and take time off for their period. Its widespread use in Spain, however, may be more linked to the fact that sex workers usually earn more when they are new to a particular place and that constant movement seems to be a strategy to bypass police checks [48]. 29 This is for instance, the case of the 2005-2006 report BEstudio sobre la explotación sexual de las mujeres, con referencia al tráfico illegal^produced by the Spanish Ministry of Labour and Social Issues which claims that there are approximately 90,000 victims of trafficking for sexual exploitation in Spain. This figure comes from considering that every sex worker who is not completely autonomous is a victim of trafficking ([23]: 22). 30 Foreign sex workers who actually claimed to have been exploited represented, in most years, less than 1.5 % of the foreign women which were identified by the Guardia Civil as potential victims of trafficking. In some years (2003 to 2005, 2009) the percentage of Spanish sex workers who claimed to have been exploited was actually higher than that of migrant sex workers. This challenges the pervading assumption, reinforced by the Guardia, that only migrants from developing countries can be and are being exploited. Source: Brazilian National Justice Council (CNJ) as quoted in the Government's official trafficking data publications [37,38] a A different data set was released by the National Justice Council [9] after the publication of the National Trafficking Report, giving larger numbers for the 2008-2011 period and establishing nine identified cases in 2012. The Council's data collection is claimed as the basis of the National Report and no explanation is given for the different numbers. It is likely that they were caused by collecting data from the fragmented and unreliable judicial sources at different points in time ([32]: 17), the Guardia admonishes against taking these numbers as the totality of trafficking cases, as it claims that for a variety of reasons (including fear of traffickers and of the police), some victims will refuse to acknowledge they have been trafficked. Although this assertion is certainly true, one cannot forget that some migrant sex workers can also, for a variety of reasons (most particularly fear of being deported), claim to be trafficking victims when they would otherwise reject the label should it not provide them with any advantages. The data from row III comes from the Ministry of Interior and refers to cases under article 188 of the Spanish criminal code (Bcoercion into prostitution^), which since 1999 specifically recognised human trafficking for the purpose of sexual exploitation as a subset. Data from the Ministry is also problematic in some other respects. First, since all crimes are transcribed into a standard data-sheet, 31 specific information related to trafficking offenses is not collected, which may make it harder for third parties to confirm whether they consist of bona fide cases. Secondly, the numbers available refer to cases of Btrafficking^which were being investigated and may or may not have end up being confirmed.
Row IV's data comes from the Fiscal Ministry through annual reports of the Attorney General's Office. Data available from 2008 to 2011 covered the more generic term Bvictims of sexual exploitation,^and included crimes persecuted under articles 31 See ( [55]: [17][18][19][20] for a reproduction of the data-sheet used by the Ministry of Interior to collect crime data. 318 bis and 188 of the Spanish Penal code detected not only by the National Police and the Civil Guard, but also by other secondary sources. 32 In its 2009 annual report (which covered data from 2008) the Fiscalía General del Estado acknowledged this problem: While a system like the current one is in force, it is extraordinarily difficult to obtain precise information about human trafficking in Spain such as it is demanded by a rational and pertinent criminal policy which aims to end this execrable crime. No official statistic can facilitate any data because the legal references do not exist. In this way, the summary report of the Attorney General's Office can reflect the number of cases initiated in persecution of the facts typified in articles 318 bis of the Penal Code (PC), 188 PC, 312 PC or any other connected to them but we would have to scrutinize each one of the procedures to learn if the pursued facts constitute in real cases of human trafficking ( [17]: 77).
The Fiscalía clarified that this problem is not exclusive to the data compiled by the Fiscal Ministry. The Ministry of Interior and the Guardia Civil also use broad definitions of human trafficking which often mix it with irregular migration issues.
Just as in the case of Brazil, different definitions are used by different Spanish institutions according to their focus area and needs. The fact that the Ministry of Interior mixes trafficking for sexual exploitation and people being coerced into prostitution and the Fiscal Ministry focuses on all victims of commercial sex exploitation in the country is easily explained when we consider that prior to 2010 Spanish trafficking legislation was problematic and often failed to draw a line between migrant sex work and trafficking. The case of the Guardia Civil, however, is much more complex.
The Guardia presents two different sets of data: one which involves all (migrant) sex workers and one which only considers those that claim to have been exploited. Although it claims this dual set of data is necessary because of its understanding that (almost) all migrant sex workers are, to some extent, being coerced, the reality may reflect a different concern. Based on her interviews with members of Spain's police forces, Malgesini ([36]: 70) stated that one of the possible explanations for the Guardia's position is that their real goal is to obtain statistics about sex work, not trafficking. Since there can be no legal registration of prostitutes under Spain's current legislation, they are labelled by the police as Btrafficking victims^so they can be tracked. This explanation also meshes with the fact that the Guardia's inspections to Brescue^trafficking victims are more often than not raids to identify and expel migrants selling sex [50].

Internal inconsistencies in Portuguese data regarding Brazilian trafficking victims
When talking about Portuguese human trafficking data, two moments must be taken into account: pre-and post-2008. Before the 2007 reform which made Portuguese 32 Although 2007 data is available (295 victims) [16], it is not included since it is limited, unlike in subsequent years, to victims detected by the Guardia Civil and the Cuerpo Nacional de Policía. Numbers from 2012 (29 victims) onwards are also not included because the Fiscal Ministry abandoned the term Bvictims of sexual exploitation^and started to address Bvictims of trafficking for sexual exploitation,^focusing for the first time on victims detected under article 177 bis [21], which, although mentioned in the previous report [20], was not then included in the final count of cases which reached the courts. trafficking legislation compatible with the UN Protocol definition, the crimes of human trafficking (restricted to sexual exploitation) and lenocínio (profiting from the prostitution of others) were considered to be one and the same.
Besides not being a true reflection of the trafficking situation in Portugal, pre-2008 data regarding trafficking victims was also not systematically available to the public. Data which specified the country of nationality of the victims was even rarer. In a similar vein to what happened in Brazil, the data collection focus in the early years seemed to be on the number of perpetrators caught, the number of police investigations that took place and the number of cases that reached the courts rather than on the victims. According to one of the few sources where such early trafficking data from the Judicial Police is available (  33 and in a special report analysing the year's trafficking statistics [11] which preceded by a year the creation of the Annual Trafficking Report. The latter is produced by the Portuguese Human Trafficking Observatory (OTSH) which was created after the 2007 change of the Portuguese trafficking legislation.
Most information about potential trafficking cases is currently collected through two standardised reporting forms: the GUR (Sole Registering Guide) 34 and the GS (Flagging Guide). GUR forms are used by the police forces (the National Republican Guard, the Judiciary Police and the Public Security Police) and the Serviço de Estrangeiros e Fronteiras (SEF, the organ responsible for dealing with foreign nationals on Portuguese soil) and represent the majority of potential cases detected. GS forms are used by NGOs and any public organs besides the police forces and the SEF. 35 By utilizing standard reporting forms and making every effort to eliminate the possibility of double counts, the Portuguese government ensures that the data which is collected is as consistent and as accurate as possible. Unfortunately, the standard forms are not always correctly or at all filled in every case, which undermines the effort. Moreover, although GS forms were created to enable the input of NGOs in the official statistics, they do little to contribute to the official understanding of trafficking in Portugal. First, because NGO-flagged cases are considered apart from those that come through GURs, as if they were somehow less valid. Furthermore, a large number of NGOs refuse to fill out GS forms, as in their understanding bringing trafficking victims to the attention of the state could be detrimental rather than helpful to their well-being [56]. 36 33 The RASI is a publication of the Portuguese Ministry of Internal Administration which details crime trends in the country. 34 A sample copy of the GUR can be found in ( [49]: 164). 35 Besides the GUR and GS forms, additional data comes from the Ministry of Internal Administration, which provides information about potential Portuguese victims of trafficking detected abroad. 36 In their analysis of the trafficking situation in Portugal and particularly the country's data gathering system, Santos Neves and Pedra [56] claim that the number of trafficking victims in the country is at least three times greater than that which is presented in official statistics.
Portuguese trafficking data is currently released twice a year. First, it continues to be published in a section of the RASI. Some months later, it is released in an expanded form in the Annual Human Trafficking Report, which has been produced by the OTSH since 2009. Although these publically available reports contain a large amount of data, they do not include all data which is collected about (potential) victims and at times specific pieces of information are explicitly redacted due to the need for secrecy in ongoing judicial and investigative proceedings and personal data protection guidelines.
Although it also contains data about perpetrators, the Annual Report is very clearly victim-centred. Per the terms established in the Portuguese trafficking victim identification guidelines, victims are divided into two categories: flagged victims (those that show enough trafficking indicators to warrant an investigation about their circumstances) and victims which have been confirmed as being trafficked.
Although the flagging procedure usually occurs upon first contact with a potential victim, the confirmation process is time-consuming and on-going. This accounts for the disparity which is found between the data presented in the RASI (which is compiled first) and that which is presented in the Annual Trafficking Report, as during the time which passes between reports some outstanding flagged cases are either confirmed or dismissed. This trend is not limited to reports from the same year; there are often cases from a given year which are still under scrutiny by the time the annual report is released and will be marked as confirmed or not in the following year's report.
Despite the internal inconsistencies of the available data across different reports due to the constant evolution of flagged cases into confirmed or not confirmed, the cases which are confirmed fulfil for the most part the definition of trafficking established in Portugal's UN-compliant legislation. However, although the Trafficking Observatory's report provides the age-range of the potential and confirmed victims, it tends to conflate in its statistics both minors and adults.
Analysing the differences between flagged and confirmed victims (Table 12) can lead to a number of conclusions. First, either the number of cases involving Brazilians has been in sharp decline over this period or the system has become more discerning in identifying suspected cases. Another possible explanation for the decline is that the number of Brazilians found in suspect situations may have decreased in the past few years due to the prevalent economic crisis. Secondly, although the cases of potential Brazilian victims represented a significant percentage of the total of flagged cases in Portugal from 2008 to 2012, 37 the actual number of confirmed victims was small. This puts in question the idea that Portugal is a significant destination of Brazilian victims of trafficking, even if it has proven to be a destination of Brazilian migrant sex workers.

Cross-comparative inconsistencies regarding Brazilian victims of human trafficking
Cross-national data comparisons, particularly in the criminal justice field, are not easily made. Differences in offence definitions, reporting and recording practices and the existence of (partially) confidential databases all lead to situations where obtaining data which is held by the same standards is difficult. If on top of that we consider the issues with internal inconsistency of data regarding trafficking victims which are present to a smaller or larger degree in all three countries being discussed, it is not surprising that the data inconsistency issue is also present when cross-comparing available statistics.
In 2013 Eurostat released a working paper on BTrafficking in human beings^ [13] which compiled trafficking data from European Union Member States and other countries 38 for the period of 2008 to 2010. It was followed in 2014 by a paper covering data from 2010 to 2012 [14]. Although the definition of trafficking used in the reports is the one found in the UN Trafficking Protocol, as well as the Council of Europe's Trafficking Convention and Directive 2011/36/EU, this definition was not always the one used by individual states when collecting data. Eurostat therefore acknowledged in the 2013 report that the results B[did] not entirely comply with the stringent requirements of the European Statistics Code of Practice^( [13]: 9) and that thus the use and interpretation of the numbers presented should be done with caution.
Despite these issues, the Eurostat compiled data will be used here as it Bpresents the first report[s] at the EU level on statistics on trafficking in human beings^( [13]: 9) In the same vein, the National Reports about Human Trafficking will be used as the data source from Brazil despite its methodological issues, as they represent the first official data compilations produced by the Brazilian government.
The Brazilian report offers yearly data disaggregated by country, while the EU report lists the specific yearly number of victims from the top ten non-EU countries of origin (in the 2008-2012 period Brazil ranks as number four, two, two, three and five respectively). As such, it was possible to compile the following table (Table 13) regarding the number of Brazilian trafficking victims identified in the selected 33 countries during the 2008-2013 period.
It is abundantly clear that the numbers are very different, with Eurostat numbers being consistently larger than the ones identified by Brazil. A number of factors explain this disparity. As the Brazilian numbers come from diplomatic representations of Brazil abroad, the fact that Brazil does not have representations in eight of the 33 relevant 38 The reports included data from what are the now the 28 EU Member States, as well as data from selected Schengen Associated Countries and candidate countries: Norway, Montenegro, Serbia, Switzerland and Turkey.  [43][44][45][46][47] *As the data of the annual trafficking reports is explicitly stated as being more accurate than that which is presented in the RASI, the data in Table 12 will come from that source when possible. 2008 data comes from the RASI since the annual trafficking report had yet to be established a Neither the 2012 RASI nor the 2012 OTSH Report state the number of Brazilian victims flagged in Portugal in that year b One or two cases of Brazilian victims were still in the process of being investigated as of the release of the report. There were also from two to six still unconfirmed cases flagged by NGOs involving Brazilians (the exact number was not given for security reasons) countries [40] is problematic. Moreover, even if a country does have one or more Brazilian consulates, it does not necessarily mean they are easy to get in contact with, particularly if people do not have sufficient financial resources and do not speak the local language. Furthermore, even if they do manage to reach the consulates, only those victims who ask for specific aid will be counted for statistical purposes, provided they are not turned away in the first place. The situation on the EU side is very different. First, we must consider that despite the existence of relevant transnational texts, not all trafficking definitions in the European States are UN-compliant, leading to different understandings of what consists in a Bvictim of trafficking.^Moreover, most trafficking victims are identified by the police, immigration officials and border guards [13]. This may mean that more victims are being identified by European states because in some countries they are being proactive in their search for them. Such is the case of Spain and Portugal, where there are periodic Binspections^in places which are suspected of harbouring victims of trafficking. On the other hand, there is no guarantee that the number of trafficking victims depicted in official statistics is indicative of the number of victims found, as these inspections/raids are often followed by the expulsion of those migrants who are irregularly in the country, without considering their situation. Whether they are acknowledged as trafficking victims at all before their expulsion may vary on a case by case basis.
Considering that the Brazilian government only takes into account a fraction of the victims identified abroad, while Eurostat aggregates data which possibly encompasses migrant sex workers and smuggled migrants, it is unsurprising that the numbers presented do not match at all. This disparity, however, is problematic, particularly if Table 13 Brazilian victims of trafficking identified in the EU and selected countries   2008  2009  2010  2011  2012   Brazilian data  48  90  77  8  6 Eurostat data a 111 151 341 b 107 89 Source: Brazilian Ministry of Foreign Relations' Consular Assistance Division as quoted in the Government's official trafficking data publications [37,38] and Eurostat [13,14] a The total number of victim identified byEurostat from 2008 to 2012 was 6309, 6955, 9710, 9438 and 10 998 b In the 2013 report [13] Brazil initially counted 76 victims and ranked as number seven. The change in the numbers, which was also significant in other countries presented in the report, is likely a result of the confirmation of cases identified in the period which occurred after the publication of the report. A similar disparity, as previously mentioned, is found between the reports produced by the Portuguese Security Ministry and its Trafficking Observatory  [37,38] and OTSH's Annual Human Trafficking Reports [43][44][45][46][47] we consider that the European Union and Brazil have made concerted efforts to establish cross-border cooperation on this matter [15]. The fact that their understanding of the dimension of the crime is so disparate can only cause problems. The disparity issues continue, to a smaller and larger extent, when comparing statistics from individual countries. Table 14 shows the number of Brazilian victims of trafficking identified in Portugal. 39 While the definitions used are assumed to be compatible, as the Brazilian Ministry of Foreign Relations uses the UN Protocol definition while the Portuguese Human Trafficking Observatory uses the current UNcompatible Portuguese trafficking legislation, the victims being counted are still different, as one group focuses on victims that sought consular assistance while the other counts those identified after an in-depth investigation.
It is interesting to note that despite constant assertions in official reports and by local trafficking specialists in both countries that Portugal is a major destination for Brazilian trafficking victims (see for instance [6] and [7]), the actual number of victims identified yearly by both sides is rather small if not non-existent. This also means that the disparities are smaller in size, if not proportion. Brazil's numbers were the higher of the two in 2009, only to have no victims identified the following year, when in turn Portugal's numbers rose. Data from 2011 to 2012 is consistent, with no victims being confirmed on either side.
Unlike Portugal (which has a central source for all of its data compilation and publication) and Brazil (which published a report with the express intention of compiling all of its existing trafficking data), Spain does not have a specific data source which can claim prominence over the others. As there is no national dataset, numbers from two of the three major available sources 40 will be compared with the Brazilian data, starting with the information provided by the Spanish Fiscal Ministry through the Annual Statistics Report of the Attorney General's Office (Table 15).
Considering that the Spanish Fiscal Ministry is detecting in this period the much more generic Bvictims of sexual exploitation,^it is not surprising that its numbers are much bigger than Brazil's in 2008, 2009 and 2011, even if victims of labour exploitation are not being taken into account. Moreover, when we consider that the Brazilian data is obtained only when trafficking victims require consular assistance, it is not unreasonable to suppose that many trafficking victims are not counted because they are unable (or unwilling, for fear of being turned to the authorities) to reach the consulates. This resonates particularly when we consider that Spain has a history of deporting trafficking victims if they are in the country irregularly, often without giving them the chance to obtain legal aid [4]. In this scenario, the fact that in 2010 Brazil detected over 75 % more victims than Spain is thus a bit surprising. 41 39 In response to an inquiry from the author, the Brazilian Ministry of Foreign Relations claimed that more specific information regarding Brazilian victims in both Portugal and Spain (such as gender of victims and types of trafficking) was not available for the 2005-2011 period (Ministério das Relações Exteriores, personal communication, August 5, 2013). Data from 2012 shows that the one victim identified in Spain (none were identified in Portugal) was a woman subjected to sexual exploitation [38]. 40 Data from the Ministry of Interior will not be used because the years where data is available do not overlap with the Brazilian data. 41 This may be accounted for by the fact that Brazil's numbers also include victims of trafficking for labour exploitation.
The problems are not limited to comparing Brazilian data with that of the Fiscal Ministry. When we take into consideration the data which is provided solely by the Guardia Civil the disparity is of an even greater magnitude (Table 16).
The reason for this disparity is easy to understand; this data set does not count trafficking victims as per the terms of the UN Protocol, but identifies instead Bpotential trafficking victims^(or rather, Bvictims of prostitution^), regardless of whether their working conditions are exploitative or not. This too broad understanding led to a situation where the Guardia Civil's numbers for 2008 and 2009 (2658 and 539), despite being its smallest figures in all listed years, were in fact far superior to Eurostat's account of the total number of Brazilian victims of all forms of trafficking detected in the European Union and selected adjoining countries in the same year (111 and 151). Despite this, these numbers produced by the Guardia are often used in academia and politics to reinforce the idea that the number of trafficking victims identified in Spain is incredibly large.
The situation becomes much more reasonable if we only take into account the Guardia's numbers on Brazilian victims who claimed to have been exploited (Table 17). Despite the fact that the data still does not match (with Spanish numbers being higher than Brazilian ones in all years but 2009), this data is at least reconcilable with that which is presented by Eurostat and can be much more easily be explained by the fact that not all victims require or desire direct consular assistance. Although not insignificant, the magnitude of the problem as assessed by both sources also puts into question the idea that there are incredibly large numbers of Brazilian trafficking victims in Spain.

Conclusion
To be able to develop an accurate response to a transnational criminal phenomenon and to keep track of how the policies being applied affect the issue, the existence of valid, internally consistence and cross-comparable data is of the utmost importance. When it Source: Brazilian Ministry of Foreign Relations' Consular Assistance Division as quoted in the Government's official trafficking data publications [37,38] and Spanish Fiscal Ministry [17][18][19][20]   Source: Brazilian Ministry of Foreign Relations' Consular Assistance Division as quoted in the Government's official trafficking data publications [37,38] and Spanish Civil Guard [30][31][32] comes to analysing the presence of Brazilian victims of trafficking in the Iberian Peninsula, this need for statistics has long been recognized. Since the late 1990s human trafficking data has been collected by all three countries and in recent years much of it has become publically available. These data collection efforts, however, have not been entirely successful. The reasons for the internal and cross-comparable inconsistency of the available statistics are multiple and often reflect some of the major pitfalls found in the collection of trafficking data worldwide. The primary problem seems to be that the definitions of Bvictim of trafficking^being used vary considerably [50]. The existence of a standard and immutable definition of human trafficking is extremely important for the development of statistics, not only for international comparison purposes, but also to enable the development of an accurate picture of the progress of the crime across the years. Mixing incompatible concepts undermines the overall data validity.
The kind of data available often depends on the institutions charged with collecting it, which have their own biases. Police and migration officers often deal with investigative data, whereas the courts tend to focus on judicial data. While such numbers may reflect the efforts made by law enforcement and legal personnel to identify trafficking situations, they often do not represent the total of identified victims and are incompatible with data from other sources which use different categories.
A second problem is that the methodology used to acquire the numbers presented, including the definitions which are being used, is often not made clear in the data collection. This means, for instance, that it may be impossible to determine whether alleged trafficking statistics are dealing with bona fide trafficking victims per the terms of the UN Protocol or if they represent, for instance, migrant sex workers who have been detected by law enforcement.
Finally, it is important to consider how the data is being obtained. In Spain and Portugal a lot of time and energy is given to police raids in clubs to detect possible victims of trafficking for commercial sexual exploitation. Besides the problems these raids bring to sex workers, this has two major consequences for data collection. By focusing on this specific venue for sex work the police and migration officers end up detecting almost exclusively women (as they tend to dominate this environment), ensuring that male and trans victims are rarely present in official statistics. Perhaps even more importantly, it leads to an under-detection of other forms of trafficking, such as labour exploitation, which are often seen as a (distant) secondary consideration, if that. 42 The idea that sexual exploitation is the primary form of trafficking taking place is thus often reproduced without any valid data to endorse it. 42 In more recent years this trend has started to change, particularly in Portugal, where greater importance has been given to detecting trafficking for labour exploitation.  Source: Brazilian Ministry of Foreign Relations' Consular Assistance Division as quoted in the Government's official trafficking data publications [37,38] and Spanish Civil Guard [30][31][32] Above all, it is important to consider that the choices which are made regarding the definitions being used, as well as data collection and processing methodology are not random or apolitical. While the collection of trafficking data may be influenced by each institution's focus area, it may also reflect systemic biases. Police forces for instanceeither because of how the trafficking legislation is worded, as is the case in Brazil, or because of how they choose to interpret the legislation, as is the case in Spainoften equate trafficking victims and migrant sex workers. Far from being an oversight, this amalgamation is quite likely to be deliberate, as it justifies the application of strong anti-sex work (and to some extent anti-migration) measures which might have been otherwise prohibited under the guise of humanitarian anti-trafficking operations. This includes raids in known prostitution locales and the denial of entry into the country of migrants who fulfil all bureaucratic requirements based on their alleged vulnerability to trafficking [50].
The fact that most of the trafficking data available in all three countries is unreliable has a number of concrete consequences. Many policy decisions made by these countries have been based on the assertion that there are large numbers of Brazilian victims of trafficking in Spain and Portugal, an assumption which is in turn contradicted or at least heavily questioned by a considerable part of the existing numbers. As it stands, none of the available statistics are particularly sound, making them unhelpful, if not harmful, to efforts to combat the phenomenon. Without accurate statistics it is impossible to establish whether the resources allocated to deal with the issue are few or disproportionate and to assess the impact of the prevention and intervention actions which have proliferated in the past decade.
More accurate information would generate a better understanding of the phenomenon and subsequently lead to more effective detection and prevention strategies and the identification of best practices to combat this crime. In the case of Brazil and the Iberian countries, a uniform understanding of the dimensions and structure of trafficking could lead to a smoother exchange of information, which is in turn crucial for effective transnational cooperation.
The implementation of such data gathering systems, while challenging, is not unfeasible. In the last decade, several attempts have been made to establish data collection frameworks which fulfil all the necessary requirements, such as the ICMPD and IOM guidelines and the MONTRASEC project. Yet besides the necessary costs (of time, money and personnel) involved in setting up and maintaining such a project, there are other considerations which could and most likely will impede (at least to some degree) the creation and/or further implementation of improved data gathering systems.
Such systems could produce more accurate trafficking statistics than the Bestimates^which are often bandied about despite the considerable criticisms they receive. But even if the more precise figures are not negligible, they do not have the same power as the overinflated numbers currently available. The mere possibility of there being thousands of victims of trafficking hidden away (particularly as they are almost always characterized as suffering from the most extreme forms of violence and exploitation) is a strong justification for maintaining the anti-trafficking policies which are currently in place in many European countries (including Spain and Portugal) and to some extent Brazil as well. The Burgency^implied in the overblown data also helps support the idea that trafficking must be stopped Bby any means necessary.^4 3 As we have seen, these policies often have much less to do with dealing with trafficking than with implementing a strong anti-immigration and anti-prostitution agenda ([24]: 22; [53]: 78). The sexist, racist, classicist and xenophobic implications of hunting down poor, non-white women 44 from developing countries at the borders and inside the EU are forgotten when these women are labelled as Bvulnerable^and the harmful actions against them presented as Bprotective measures.^As long as the supposedly huge figures of B(potential) trafficking victims^allow states to implement their real agendas behind a protective screen of alleged human rights protection, it is unlikely that the necessary changes to the existing data gathering systems will be made.

Introduction
During recent years, corruption cases connected to urban development have become an important problem in Spain. The disproportionate evolution of the Spanish housing market during 1998-2007 (6.5 million housing units were built in that period) triggered massive, disordered urbanization on pieces of land that were not always appropriate. In turn, this massive urbanization paved the way for opacity, wrong policies and corruption. Thus, instead of an urbanization that protected citizens' interests, citizens were ignored for the sake of urbanization agents' benefit.
Municipal corruption scandals have been reported not only to court, but also to other institutions such as the European Parliament Petitions Commission and the Ombudsman. Actually, the connection urban development-corruption is not new. What is new is the extent of the problem. Corruption cases reported by the press suggest that this is a systemic rather than sporadic problem. In fact, 8.3 % of Spanish municipalities faced at least one case of urban corruption in 2000-2009 [23]. Furthermore, according to the 2013 Special Corruption Eurobarometer, 63 % of respondents in Spain (highest percentage in the EU) feel personally affected by corruption in their daily lives (EU average: 26 %), while 95 % say that corruption is a widespread problem in the country (EU average: 76 %) and 91 % state that corruption exists in local and regional institutions (EU average: 77 %). Furthermore, Spanish citizens are more likely than any other EU citizen to think that corruption is widespread in both political parties and among politicians, that the level of corruption within the country has increased and that high-level corruption cases are not sufficiently punished [13].
Corruption is the abuse of entrusted power for private gain. This is the definition used by Transparency International (TI), which applies to both public and private sectors. Corruption generally comprises illegal activities, which mainly come to light only through scandals, investigations or prosecutions. In Spain, urban corruption usually consisted in businessmen bribing politicians, not government officials, to get things done. Basically the plot started with the businessman buying a cheap plot because it was non-developable. Afterwards, the politician changed the status of the plot to developable, which made its value skyrocket. Thus, instead of working towards citizens' interests, politics pursued the benefit of both businessmen and politicians involved. This kind of corruption is different from third-world, where low-paid government officials take bribes for daily bureaucracy steps.
Furthermore, political power in Spanish municipalities is highly discretionary. Mayors are more powerful if we compare with other countries, which put together with a flexible legislation, paves the way for corrupt behaviour. Spanish local governments (LG) belong to the 'strong-mayor' type [29]. The strong-mayor form, typical of the Southern European countries -France, Italy, Portugal and Spain-gives a central role to the mayor for local affairs management. They hold most executive powers in their hands, compared to local managers, officials and employees, who maintain a secondary role. As we will see in detail later, one of the most important responsibilities of the Spanish local authorities concerns urban planning and construction. These are the key aspects of the urban planning activity of Spanish municipalities [18]: -They decide what areas in their jurisdictions are fit for urban development.
-They issue building permits according to their local urban plans, after receiving technical and legal advice from the local urban planning officers. These permits are required for landowners to start the urbanization project. -Frequently, local authorities and private landowners sign development agreements to set the urban development of a private land. These agreements typically include details such as quantity and type of building, green areas, local public facilities, etc. Until 2007, the procedure for signing such agreements was barely regulated, which meant that mayors and landowners and/or urban developers could negotiate the conditions for urban development bilaterally with minimum legal requirements to report to citizens or even to the municipal corporation on the content of the agreements.
Under this legal framework, local urban development in Spain created strong economic incentives to increase the areas eligible for urban construction, and provided opportunities for corrupt exchanges between public and private actors. Given that urban policies in Spain can generate large capital gains, our analysis focuses on the determinants of Spanish municipal urban corruption during the huge 1998-2007 property bubble.
Besides, the huge number of municipalities in Spain, over 8,000, prevents the media from investigating them thoroughly. It is feasible to control the largest municipalities, but it is difficult to do it with all of them. The large number of LG with a monopoly of regulatory power increases the total number of potential bribes [39]. This high number of Spanish municipalities and their reduced size-a Roman and medieval inheritance [1]-makes it is costly to employ sufficient and competent staff and to ensure that they have the necessary independence to control all reprehensible behaviour by local representatives. We must bear in mind that Spanish local authorities are crucial holders of territorial power in Spain; however, they have a small size and limited resources. Indeed, over 90 % of local authorities' population is under 10,000, making the Spanish councils among the smallest in Western countries, with an average population of 4,800 inhabitants.
Moreover, we must highlight that it is very complicated to assess absolute levels of corruption in countries or territories on the basis of hard empirical data. There are three principal approaches to measure corruption at the macro level, namely, (1) general or target-group perception, (2) incidence of corruptive activities [35] and (3) bribes reported, the number of prosecutions brought or court cases directly linked to corruption.
The first kind of measures reflects the feeling of the public or a specific group of respondents about corruption. The second approach is based on surveys among those who potentially bribe and those whom bribes are offered. The third kind of measures groups more objective variables that are also used as indicators of corruption levels.
Golden and Picci [20] consider that survey-based measures of corruption have some intrinsic weaknesses. First, the real degree of reliability of survey information is largely unknown. Thus, respondents involved in corruption may have incentives to underreport such involvement, and those not involved typically lack accurate information. Second, the reliability of the indexes may also deteriorate over time. As the indexes are widely publicized, there is a danger that survey respondents, rather than reporting how much Breal^corruption exists, are reporting what they believe based on the highly publicized results of the most recent indexes. Furthermore, respondents may lie in surveys to improve the results of the indexes.
Many organizations, such as Transparency International (TI), European Commission, World Bank or The Heritage Foundation, report corruption indexes for different countries, and publish recommendations to curb corruption. In this respect, there are three different types of anti-corruption policies [38]: (1) Administrative reform: it is the most common approach to address corruption.
Anti-corruption strategies focus on enhancing the quality of bureaucracies, merit-based promotion and recruitment through special exams, narrow job definitions, establishment of formalized rules, accountability and responsibility. Furthermore, reformers have also tried to apply New Public Management tactics to protect public organizations from political influences. (2) Law enforcement: it is an important alternative to complement administrative reform and particularly significant in countries whose governments have little internal capacity to implement these reforms. Imposing legal constraints and prosecution of corruption makes an example to all government workers, while also helping to reinforce government employees' expected standards of behaviour. (3) Social capital: it is based on cultural norm or network that facilitates modern economies and that can enhance the rule of law by providing society with information resources, morality, trust and civic associations. Thus, citizens are more likely to be protected from political exploitation and should also act as surveillance guards against public officials' corrupt behaviours. This approach is more relevant for complex societies in which the use of formal control instruments is difficult to apply.
In a more practical approach, focusing on the particular case of municipalities, TI gives some recommendations to fight against corruption [27]: & The remuneration of city councillors should be determined according to the nature of their workload, as well as the size of the respective local government area. & There should be clear rules restricting the engagement of city councillors in private activities to avoid potential conflicts of interest. & Mayors and city/county councillors, as well heads of departments should be required to regularly disclose information related to gifts received, all properties and sources of income, debts and liabilities, shares in companies, as well as potential conflicts of interest. & In certain circumstances, municipalities should also seek to restrict pre-public employment (the movement of businesspeople to the local administration) and post-public employment to avoid undue influence and misuse of confidential information.
Similarly to the aforementioned international organizations, many studies have analysed the empirical determinants of corruption at national level using indexes that measure perceptions rather than actual political corruption. Nevertheless, there is not commonly agreed-upon theoretical approach on which to base an empirical model of corruption, let alone to investigate the causes of corruption [3].
The evaluation of corruption is even more complex at the local level. Thus, while numerous case studies in particular countries and cross-national comparative research have examined the details of corruption, municipal research is scarcer due to two factors. First, the difficulty of measuring corruption levels in LG has become a major obstacle [16]. Second, it is difficult to get reliable data.
Our paper contributes to close this perception gap. Specifically, the aim of this paper is to analyse the socio-economic and financial determinants of urban political corruption at local level through a dataset of Spanish municipalities. As we stated above, traditionally, works assessing corruption have focused mainly on cross-country datasets. However, the inability to account for the full set of institutional arrangements that determine corruption makes results difficult to interpret. Only a few studies have considered the municipal level, despite municipal datasets have two main advantages over cross-country, namely, homogeneity and amplitude [31,6]. Besides, we will use, as gauge of corruption, the third measure we have mentioned above, i.e., the number of cases of urban corruption reported in the online press in the period 2000-2009. We aim to assess the level of municipal corruption, so that later we can guess the determinants of urban corruption.
Our findings show an impact of transparency, salaries and population (as a gauge of urbanization) on corruption. First, the lower the transparency, the higher the probability of corruption. Second, higher wages ensure less corruption. Third, larger municipalities experience more corruption.
The paper is organized as follows. Section 2 discusses how the literature has analysed the relationships between corruption and socio-economic and financial features. Section 3 describes the urban planning in Spain. Section 4 addresses the research design and methodology. Section 5 presents the regressions. Section 6 discusses the empirical results. Finally, Section 7 summarizes conclusions and proposes further research.

Literature on corruption
Why do some municipal politicians misuse public office for private gain more frequently and for larger payoffs than others? The answer lies in the balance between the expected cost of a corrupt action and the expected benefit [41,37]. Recent theoretical and empirical research has considered how differences in socio-economic and financial features might explain variations in the level of corruption. In this section we present the main contributions to the literature on the determinants of corruption. Nevertheless, as we said before, most of these studies focus on national rather than municipal datasets.

Transparency
Both the provision of information to citizens and citizens' ability to monitor their governments play an important role to battle against corruption. In the last years, many governments have increased openness and transparency. Information and communication technologies (ICTs), in general, and E-government, in particular, are seen as suitable means to promote transparency and to reduce corruption [8].
Bac [7] highlights two opposing effects of transparency on corruption. First, a higher level of transparency in decision making increases the probability that corruption or wrongdoing is detected. Second, it may also help outsiders to identify key decision makers, thereby enabling connections for corruption. The Bconnections effect^may outweigh the Bdetection effect^and thus, higher transparency may boost corruption, a prediction sharply in contrast with standard theories of transparency.
Alt and Lassen [3] also consider that the effects of transparency may be ambiguous. Transparency might increase the detection of corrupt acts, or reduce corruption when the expectation of corrupt activities being observed is sufficiently internalized.
Nevertheless, a smaller, indirect effect of transparency is that increasing government scale also rises the temptation to be corrupt [4].
Lindstedt and Naurin [26] conclude that reforms focusing on increasing transparency should be accompanied by measures for strengthening citizens' capacity to act upon the available information, if we aim to curb corruption. With the above considerations in mind, we propose hypothesis 1: H 1 : The level of transparency influences the level of corruption.

Income
Income is a classical variable used to explain corruption [35]. Nevertheless, its impact on corruption is unclear. On the one hand, most authors conclude that corruption is lower in richer societies [41,3,30,21,22,12,36,26,19,37,16,10]. On the other hand, some studies deviate from this mainstream, because they show that income increases corruption [11,17]. These latter authors consider that corruption has a procyclical nature. Thus, they stress that Bmoral standards^are lowered during economic booms, as greed becomes the dominant force for economic decisions. Accordingly, we propose hypothesis 2: H 2 : The level of income influences the level of corruption.

Education
Population's education level is also argued to affect corruption. Corruption is expected to be lower when populations are more educated and literate ( [41]; Ali and Isse 2003; [30,3,19,16]). The reason is that governments are better monitored by a cultivated civil society. However, counter-intuitive findings are found in Fréchette [17], Seldadyo and de Haan [35] and Shabbir and Anwar [37], who show that when population becomes more educated, corruption increases. Their point is that, in developing countries, the public sector is the main source of employment. In these countries, corruption in public sector is very common and becoming a civil servant requires education. Therefore, the level of corruption in these countries rises with the increase in education, especially when it becomes the source of employment in the public sector. Finally, Damania et al. [12] do not find a significant relationship between education and corruption. Therefore, our hypothesis 3 stands as follows: H 3 : The level of education influences the level of corruption.

Urbanization
Some literature has identified urban development as another variable that explains corruption. According to Billger and Goel [10], urbanization may have two opposite effects on the level of corruption. On the one hand, greater concentration of the population in urban areas increases potential bribe takers and bribe givers, making them more eager Bto jump the queue^via illegal means. There are also greater opportunities for interaction between potential bribe takers and bribe givers in urban areas, resulting in more corrupt deals. On the other hand, a highly concentrated urban population might indicate a greater chance that someone is looking over the shoulder of potential bribe takers and bribe givers, acting as a deterrent.
Empirical findings are not conclusive about urbanization either. Meier and Holbrook [28] and Alt and Lassen [3] show a positive relationship between urban concentration and corruption level. However, Hill [22] and Billger and Goel [10] find the opposite result, i.e., there is more effective government oversight in urban communities or potential bribe takers and bribe givers are somewhat deterred by peer pressure in urban areas. Finally, other studies do not find a significant relationship [12,19,16]. Therefore, we propose the following hypothesis: H 4 : The level of urban concentration influences the level of corruption.

Debt and treasury surplus
The increase of funding resources, such as debt and treasury surplus, may be also related to corruption.
Corruption is defined as the misuse of entrusted authority for private benefit, therefore, governments with higher funding may have a higher probability of corruption by using this money [41]. In this way, the literature supports that lower corruption is associated with lower opportunistic public debt and treasury surplus. Nevertheless, governments with higher levels of debt may be monitored by lenders to a greater extent. In this case, a negative relationship between the level of external debt and corruption is expected [35]. Accordingly, we present the following two hypotheses:

Salaries
The literature has developed the theory of how Bcontrol^can curb corruption [11]. We refer to a sort of models where the ability to keep bureaucrats under control depends on the auditing intensity and the combination of sticks (fines, dismissal) and carrots (salary, prestige, pensions) offered to the agent. In this context, Van Rijckeghem and Weder [42] argue that civil servants' wages may be important. Public sector wages are highly correlated with the measures of the rule of law and the quality of the bureaucracy, and therefore may have an effect on corruption. Thus, if bureaucrats receive low wages, they have incentives to be corrupt.
There are many empirical studies that link corruption (and the temptation to act corruptly) to the level of salaries in public office [41]. Van Rijckeghem and Weder [42], Herzfeld and Weiss [21], Alt and Lassen [3] and Beylis et al. [9] find evidence of a significant relationship between relative civil-service pay and corruption. The relationship implies that a rather large increase in wages is required to eradicate corruption solely by raising wages. However, Seldadyo and de Haan [35] show a counter-intuitive result, namely, that an increase in government wages lifts corruption up. Therefore, we propose the following hypothesis: H 7 : The greater the salaries of politicians, the lesser the level of corruption.

Urban planning in Spain
The Spanish public sector is divided into three levels: the State, Regional Governments (called Autonomous Communities, 17) and LG (50 provinces and 8,117 municipalities). The relationship among them is based upon competences, not upon hierarchy.
Each municipality has a mayor, a cabinet, and a professional administration. The mayor is the head of the executive, and is elected indirectly by the citizens. The electoral system is based on proportional representation. Municipal governments are elected once every 4 years.
One of the main responsibilities of LG is to elaborate and execute land-use plans. The whole municipal territory is divided into three land categories: non-developable, developable and existing urban land. The principal purpose of local land-use plans is to decide about the non-developable or developable land. The legal town planning framework has been based on three basic essentials: (1) all land in the country was 'classified' by municipal plans as fit or unfit for building and urban development; (2) most of the capital gains generated by land classification were rendered to the owners of the land regarded as fit for development, and just a small part of it (10 to 15 %) was recovered by the public administration that decided on land use; and (3) if a public administration needed to expropriate land for public use, the law required a valuation that prevented in practice to expropriate land classified as fit for urban development.
Taking into account these three elements, while rural land was regarded unsuitable for building and development (with no right to claim any compensation), the land that the municipal plan classified as fit for development gained a totally different legal (and economic) status. Moreover, and in the case of expropriation, rural land that became urban land under the municipal plan would be valued as if already was fully developed (urbanised and built on) simply by the plan being approved. Of course, this singular characteristic usually became an extraordinary source of speculation and corruption. This absolutely irrational element favouring landowners' interests made irrelevant the threat of expropriation whenever landowners did not comply with the time limit to develop their land in accordance with the municipal plan, allowing huge land speculation: landowners held their plots without investing a euro in urbanising works, waiting for an increase in land price.
Besides this faulty institutional design, from mid-1980s onwards a new significant element in urban planning emerged without any supporting legislation. Some important city councils started to sign urban planning agreements with developers who would be willing to fulfil more commitments for the council than required by law, in exchange for amendments in the current urban plan, including the rezoning of some plots. Although these agreements were well intentioned (for instance, to guarantee that developers would build free housing for the homeless, in exchange for permission to develop slum areas), its opaque nature led soon to corruption scandals. The agreements were signed just by the developers and the town mayor and allowed amendments of the existing urban plan -in a legal procedure with little publicity and citizen participation-which very often meant a complete transformation of the city model envisaged in the amended plan. The extraordinarily widespread use of these urban agreements in Spanish LG speaks of the greater flexibility in the approach to urban development, but also explains the growing problems of corruption in this field.
TI developed a procedure to help curb corruption at the municipal level, BThe Local Integrity System (LIS) Tool^, which assesses the internal governance and capacity of each of the core local government actors and their role in promoting integrity in the system as a whole. It also assesses the capacity to perform, and the effectiveness of, each of the oversight and accountability functions. When used repeatedly, the LIS Tool can monitor and evaluate the progress of the LIS over time. If the LIS is not well defined and implemented, and if the mayor has a strong position, both facts may lead to the adoption of short-sighted public policies and to a very serious level of local corruption, which has been both effect and cause of the property bubble [33]. Under the Spanish legal framework, urban planning discipline rests mainly in the hands of town councils, particularly of the mayor, who issues building permits that a developer must obtain before starting any urbanisation activity. These permits must be issued in accordance with the current urban plans and after receiving technical and legal advice from municipal officers.
In addition to the countless corruption cases triggered by urban development in recent years (Section 4.1 shows the corruption practices that Spanish LG use more frequently), LG have undoubtedly used this activity to obtain a major source of funding. Specifically, urban development decisions have an impact on several local revenue sources. One of the most important revenue sources comes from the free transfer of a percentage of the building rights from landowners to local authorities. As we said before, Spanish laws establish that some portion of the benefits accrued from urban development must go to citizens, rather than being captured exclusively by the landowner. In particular, landowners must cede a portion (10-15 % in general) of their building rights to the municipality, either in the form of land or money. While the ceding of land does not appear in the budget, the amounts obtained from the cash payments do. Nevertheless, LG can sell the lands previously ceded by landowners, which involves an increase in the budgetary revenue.
Spanish LG also levy funds from other revenue sources related to urban development. There are two local taxes directly linked to urban development activity: the land disposal tax and the construction tax. The first tax is paid by the seller of urban or developable land. The construction tax is paid by individuals or companies who engage in construction projects (new buildings or renovation works). Furthermore, LG impose fees on developers to finance the capital costs of additional public works and facilities that are necessary for an appropriate land development. Municipalities also gain revenue from fees for planning permission. In addition, LG receive funds from the granting of use rights of municipal-owned property assets. Consequently, as urban development increases in the municipality, the money generated by these revenue sources also increases. This makes LG an interested party in the urban land development [24].
The local property tax is another source of revenue for Spanish municipalities that may appear to be related to urban development. However, contrary to the construction and land transactions taxes, the property tax does not depend on the dynamics of urban activity. In fact, it can be considered as a recurrent revenue source, for it depends on the stock of dwellings.
It is important to point out that under Spanish laws, some of these revenues must be devoted to the promotion of social housing and other social purposes. In particular, the funds received from landowners for the donation of a part of their building rights and those obtained from the sale of lands previously ceded by landowners, must be devoted to these purposes. Since the law provides that these resources can be used for other social purposes than the promotion of social housing, LG have widely used these revenues for the construction of public facilities, such as sports centres, theatres and parks. In other words, these resources can be used by LG to finance capital expenditure.
To summarize, LG have several sources of revenue that are related to urban development. Essentially, we can distinguish (i) the money paid by landowners for the donation of a part of their building rights, (ii) revenue from the sale of lands previously ceded by landowners, (iii) the land disposal tax, (iv) the construction tax, (v) fees on developers, (vi) fees for planning permission, and (vii) the funds received from the granting of use rights of municipal-owned property assets. Henceforth, when we refer to municipal urban development revenue, this is assumed to mean the sum of all these revenues with the exception of revenue from land sales. Unfortunately, this information is not available for most municipalities.
In Spain, the recent housing bubble contributed to the rise in municipal urban development revenue, while the bubble burst made these revenues plummet. Figure 1 shows the evolution of municipal urban development revenue over the period 2003-2010. This period includes the peak years of the Spanish housing boom and the early years of the burst. The evolution of two housing market indicators (house prices and building permits) is also presented in Fig. 1 to allow comparison. In Spain, house prices increased markedly until 2007, after which they plummeted. The escalation of house prices before the burst in 2008 triggered a spectacular construction boom, as shown by the sharp increase in building permits. The subsequent fall in house prices was accompanied by corresponding drastic declines in building permits. Municipal urban development revenue followed a similar trend, with an enormous growth until 2006, while in 2007 it began to decline. However, it was in 2008 when urban development revenue plummeted. Therefore, as it can be observed, the Spanish housing boom made urban development revenue skyrocket in LG, while the collapse of the housing market led to the burst of the revenue boom.
Finally, it is important to point out that prominent corruption cases investigated in the recent years have raised awareness of potential corruption risks and have increased public authorities' focus on the need to strengthen anti-corruption and integrity-related policies. The central government has acknowledged the need to address corruption as a matter of priority, along with other economic recovery measures (European Commision 2014b).
The need to curb corruption has triggered other countries' initiatives. For example, Germany has introduced some measures to fight against corruption at the municipal level. These policies include rotation of staff, strict observance of the 'four eyes' rules; clear regulations on sponsoring and the prohibition on accepting gifts; establishing centralised authorities for tender/awarding; precise description of the tender and control of estimates; organisation of tender procedures, including secrecy of bids and prevention of belated manipulation of bids; increased use of e-procurement; documentation of adjudication and careful control by supervisory bodies; exclusion of enterprises found guilty of corruption offences and establishing black lists/corruption registers. In Italy, several networks and associations of regional and local administrations are actively implementing actions for prevention of mafia infiltration in public structures and promoting transparency of public procurement at local level [14].
In the Netherlands, most cities and communities have developed a local integrity policy, and some of them have created integrity offices (IO), which support all municipal units with the following services: advice, training, risk assessments, handling of disciplinary cases and legal advice and integrity investigation. IO acts as a contact point for people who want to report a breach of integrity rules. It also advises businesses, service providers and even other municipalities in the identification, control and reduction of integrity risks. If a department or service proposes a penalty to be imposed for a breach of integrity rules, it has to request an opinion from the IO's inhouse lawyers. The purpose of this mandatory request is to ensure that the policy on penalties for such breaches is consistent [14].
Considering this, political corruption is an issue that requires further analysis. This is particularly important for big LG, which are included in our sample, since they manage huge amounts of financial resources. In spite of the high number of investigations into allegations of corruption and the shortcomings mentioned above, no comprehensive approach has been developed as a basis for addressing particular risks and vulnerabilities at regional and local levels [15].
According to Pettersson-Lidbom [31] and Ashworth et al. [6], as we said above, municipal datasets have two main advantages over cross-country, namely, homogeneity and amplitude. First, heterogeneity (different legal structures and socioeconomic framework in cross-country samples) needs to be controlled [40]. This feature is overcome in the municipal level within a country. Second, sub-national datasets are considerably larger than cross-country samples.

Methodology Variables
The variables included in our model stem from the theoretical and empirical underpinnings (Table 1 depicts descriptive statistics and expected signs).
Our dependent variable is the number of cases of urban political corruption detected in Spanish LG for 2000-2009 (corruption). The property bubble period started in 2000 and in 2008 it burst. That is the reason for taking this time window plus 2009 because it is reasonable to think that some corruption cases connected to the aforementioned property bubble could occur after that period. This information comes from the dataset made by professors Jerez, Martín and Pérez (University of La Laguna, Spain). Their database was constructed for the research project BLand urbanization and local policy in Spanish democracy: an insight on agriculture^funded by the Spanish Ministry of Science and Education. Due to the lack of official statistics on municipal corruption, these authors used online press to identify urban related corruption cases for 2000-2008.
After a thorough research of corruption press articles, they came out with a dataset of 414 cases. This figure shows clearly the importance of the problem and the peculiarities of the Spanish urban system. Nevertheless, as we said above, corruption cases stem- Corruption cases included in our variable are summarized on Table 2. We must point out that politicians entitled to make and enforce the laws in the people's name, are using this authority to sustain their power, status and wealth. Political corruption not only leads to the misallocation of resources, but it also perverts the manner in which decisions are made. Political corruption occurs when the laws and regulations are abused by the rulers, side-stepped, ignored, or even tailored to fit their interests. It is when the legal bases, against which corrupt practices are usually evaluated and judged, are weak and furthermore subject to downright encroachment by the rulers [5].
According to the previous literature, we take the following socio-economic and financial features as explanatory variables: transparency index (transparency), the mean of income level of the municipality (income), the rate of uneducated people in the municipality (uneducated), the rate of urban revenues on total revenues (urbanrev), the size of the municipality (population), the rate of municipal debt growth (debt), the rate of municipal treasury surplus growth (treasurplus) and the ratio of LG politicians' salaries over non-financial expenditures (salaries).  The implementation of these indexes aims to achieve two goals. First, each local government gets an individual transparency score, so that a transparency ranking is published. Second, it attempts to promote the culture of full disclosure, as it offers the municipalities the opportunity to publish the requested information in the municipal webs, and this way improve their transparency scores.
The index aims to foster a closer relationship between councils and citizens, encouraging the increase of information disclosed about the situation of the municipality and the activities it carries out.
In May 2009, local governments received a questionnaire with 80 items. Some information was directly collected by TI-Spain. The remaining information was provided by the municipalities through the questionnaire. The participants had to indicate, in an appendix, the source of each data, so that TI-Spain could check it. The nonresponding municipalities will be assigned a minimum score (out of the 110 initial Councils, 18 did not return the questionnaire, and therefore received the minimum score).
Each question had two possible scores: a. 2 points: if the information is posted on the municipal website. b. 0 points: If the information is not published on the municipal website. • Classification of non-developable plots as developable plots in lands that due to their natural features, should be kept as non-developable. • Forbidden building up of non-developable land, either with the municipal approval or taking advantage of the municipality's negligent inactivity on urban illegalities control. • Classification as urban developable of land that should not be classified according to the law.
• Partial modification of the municipal urban plan instead of a general revision of the plan, which would be more appropriate. This partial modifications distort the general urban plan of the municipality. • Illegal urban activities are legalized through ad hoc plan modifications, which prevents the judicial sentence from being executed. • Urban plans or urban plan modifications are approved without meeting urban laws requirements.
• Urban developments are executed circumventing some basic legal requirements • Municipal land disposal revenues, which should be reinvested on municipal land, sometimes finance municipal current expenditures. • Municipal land not used for the required legal purposes.
• Urban plans ignore environmental legislation, which causes ecological damages.
Finally, depending on the total value obtained by each participant (the sum of their scores in the 80 indicators), TI-Spain developed the ranking.
These 110 municipalities make our sample, since the variable transparency constrains our analysis to those LG included in the TI-Spain survey.
Urbanization is measured through two variables. First, the size of the municipality (population), which is a gauge of the urban development of the city. Second, urbanrev, which is the sum of revenue sources directly related to urban development decisions as we define it in Section 3, that is: (1) money paid by landowners for the donation of a part of their building rights, (2) the land disposal tax, (3) the construction tax, (4) fees on developers, (5) fees for planning permission and (6) funds received from granting use rights of municipal-owned property assets. Property tax collection has not been included because it depends on the stock of dwellings.
Regarding salaries, our independent variable captures top politicians' wages as an aggregate figure for the wages paid to all top politicians. In this respect, top politicians are defined as the members of the municipal council that make up the municipal government. In our view, it is more accurate to use the ratio of all top politicians' wages over non-financial expenditures because it enables the comparison among the LG considered in our sample.

Specification of the model
The initial specification of the model corresponds to the following equation: corruption=β 0 +β 1 transparency+β 2 ·income+β 3 ·uneducated+β 4 ·urbanrev +β 5 · population+β 6 ·debt+β 7 ·treasurplus+β 8 ·salaries+ε We analyse two different approaches: Ordered Logit Model (OLM) and Ordinary Least Squares (OLS). The former model is more appropriate than OLS in our analysis due to the ordinal nature of the dependent variable. As we defined above, our dependent variable is the number of cases of corruption for the period 2000-2009; therefore, it is discrete and ordered, inasmuch as less cases of corruption (or none) are better than more cases.
Appropriate for the modelling of an ordered categorical dependent variable are both ordered probit model and the ordered logit model. In these models, the probabilities of each outcome, conditional on the independent variables, are modelled using the cumulative normal distribution or Weibull distribution, respectively (Collet 2003). In case the residuals are not normally distributed, and ordered logit model should be applied. This is the case for the data at hand, thus, we apply the ordered logit model.
We also use White-corrected OLS method that ignores the categorical nature of the dependent variable both to identify variables that clearly affect the level of corruption and to check the robustness of our results [2].
At the outset, we considered both uneducated and income as independent variables, as suggested by the literature. Nevertheless, these two variables present a high correlation in our sample: Pearson=−.659, significance=.000. Therefore, we dropped uneducated in order to avoid multicollinearity problems.

Sample
The sample consists of the 110 largest Spanish municipalities surveyed by TI-Spain. The time window covers 2000-2009. Municipalities' population ranges between 35,396 and 3,255,944 in 2009. This sample represents nearly 48 % of the population. The reason for eliminating the rest of municipalities is twofold. First, the reliability of the financial data is doubtful for small LG. Second, a key variable for this study, transparency, is only available for these municipalities, as we explained above.
We think that LG in Spain provide an ideal institutional setting to test our hypothesis for several reasons. First, as we said above, Spain is perceived as being among the most corrupt countries in the European Union [13]. Prominent corruption cases investigated in recent years have raised awareness of potential corruption risks and increased public authorities focus on the need to strengthen anti-corruption and integrity-related policies. The national government has acknowledged the need to address corruption as a matter of priority, along with other economic recovery measures. Second, LG seem to be the most corrupt in many countries.
LG are under the control of narrow elites that use several tools for personal gain [34,16]. Spain is one of the most decentralized countries in Europe, and it clearly shows the importance of sub-national government's control of corruption. Finally, executive accountability is limited. The Spanish Supreme Audit Institution (SAI) is the only body responsible for auditing government accounts and financial management. Although the agency is legally independent, in practice it is influenced by the two major national political parties. The institution has sufficient resources, but is not very effective in controlling public sector's efficiency and effectiveness. Table 3 shows the coefficients of the OLM and OLS regressions. There are only slight differences in the coefficients of both estimations, which confirms the robustness of the model. Overall, OLM coefficients present more significance than OLS coefficients. Anyway, in both regressions, the values are significant according to the usual statistical threshold of p<.05.

Discussion
The empirical results give support to our hypothesis 1, since higher transparency reduces corruption (transparency). Our results show that, in fact, one of the measures suggested by TI to curb corruption, i.e., enhancing public institutions' transparency, plays a key role to curb political corruption. Lack of transparency prevents political opposition and citizens from supervising urban development decisions made by the municipal body. This supervision is usually difficult to implement in Spain because a) it is a task that requires a lot of time and effort, and no all municipal council members are willing to do it, b) these members usually lack urban legislation knowledge, c) the municipal legal department is reluctant to provide full information to the opposition, d) the mayor refuses in many cases to provide the information the opposition requires and e) sometimes opposition lacks financial resources to take other politicians to court. Besides, urban planning agreements with developers, that were discussed in Section 3, are signed by the mayor without the approval of the municipal council, which hinders opposition's control.
In light of Spanish citizens' deep concern about corruption that appears in all opinion polls, the Spanish Parliament passed in February 2013 a resolution called Transparency Act. This act lays down a wide range of anti-corruption measures [15]. Against this background, an ambitious programme of legislative reform is being implemented, covering a wide range of aspects, including financing of political parties, criminal law aspects, accountability of high-ranking officials and increasing the efficiency of court proceedings. Furthermore, in September 2013, the Government approved a package of measures for the 'revival of democracy'. The package comprises: strengthening SAI powers; clear rules on the obligations attached to exercising public office and the corresponding sanctions in case corrupt behaviour is detected and a proposal to reform criminal law and criminal procedure rules, including the sanctioning regime applicable to corruption offences.
In respect of salaries (salaries), the data prove our hypothesis 7, for higher salaries mean lower corruption. This is in line with Van Rijckeghem and Weder [42], who suggest that high wages discourage politicians to be corrupt. As we pointed out in the literature review, while there is a growing range of studies on the link between wages and corruption, findings on whether higher salaries reduce incentives for corruption are mixed. Some studies conclude that anti-corruption policies designed to increase wages and net income of potentially corrupt agents may be ineffective. Moreover, La Porta et al. [25], contrary to our study, even found that higher government wages are correlated with more corruption. There is an emerging consensus that increasing salary may not be sufficient to curb corruption, in the absence of effective controls and management of staff and resources.
However, we think that one of the main arguments on the link between low salaries and corruption is that for top politicians with low salaries, corruption becomes a coping strategy to compensate for economic hardship. This Bneed-based^argument stems  from underpaid politician accepting bribes for basic necessities [32], as opposed to Bgreed-based^corruption, which is more apparent in cases of well-paid politicians in higher level positions [43]. Urbanization is connected with higher corruption, as the coefficient of population indicates on Table 3 (hypothesis 4). Though the literature is not conclusive in this respect, our results are in line with Meier and Holbrook [28] and Alt and Lassen [3], who defend a positive impact of urban concentration on corruption.
Finally, all other control variables turned out to be non significant as determinants of corruption (hypotheses 2, 3, 5 and 6).

Conclusions and further research
This paper evaluates the impact of socio-economic and financial factors on urban political corruption through a sample of the 110 Spanish largest municipalities for 2000-2009. This is a topical issue in Spain. For example, in the moment these lines are written (24 April 2014), the front page of one of the main Spanish newspapers (El País) reads: BThe Spanish Attorney-General claims that more resources and laws are needed to curb corruption^. In this context, our findings indicate that measures to stop corruption should include an adequate salary scheme for politicians and full transparency on governments' financial decisions and reports. However, the financial crisis in Spain has led to Spanish Parliament to adopt a measure consisting to limit the salary of the mayors, which currently has no legal limit. In the light of our results, we think that mayors should be well-paid and a salaries scheme for all municipal politicians should be set, as a way to curb corruption at the municipal level. A discussion, therefore, is opened about what we must understand for being well-paid; in this way, the discussion has to be responsible and free of demagogies. Our results confirm TI's suggestion regarding municipal politicians' remuneration, i.e., their salaries should be set according to their workload, as well as the size of the respective local government area.
Transparency of municipal reports should also be enhanced, as TI-Spain claims, as a way to enhance citizens and opposition control over government. We would like to emphasize the problem of urban planning agreements, which should be completely reformed by the Spanish Parliament. These agreements are opaque and foster the agency problem between principals (citizens) and agents (politicians). Besides, too many decisions are at the discretion of the mayor, which paves the way for corruption practices. However, increasing transparency is not enough, but it should be accompanied by measures for strengthening citizens' capacity to act upon the available information. In this respect, transparency should go beyond budget reports, and as TI indicates, local politicians should be required to regularly disclose information related to gifts received, all properties and sources of income, debts and liabilities, shares in companies, as well as potential conflicts of interest.
Regarding municipal population, our results suggest that a different legal treatment should be applied to the largest municipalities. At present, in Spain, these cities have specific legislation regarding revenues and competences. This ad hoc regulation should also include measures to prevent corruption from happening in these big cities, such as tighter control from the SAI, closer budget and urban surveillance from the regional and/or central government, more citizens' participation channels to enhance democracy, etc. Some authors have even claimed that urbanization competences, currently devolved to municipalities, should return to the central government.
As far as further research is concerned, new datasets should improve the information about corruption. For example, taking information from courts' sentences on corruption cases. This would allow us to complement our analysis based on newspapers information and to incorporate the amount of money stolen to the dependent variable. Furthermore, institutional features that impact municipal corruption could be investigated in an international comparative approach.
contextualizes prostitution in the country's unequal economic development and the consequent population migration. She gives an outline of Chinese prostitution history and points out that it is a revival and has now reached its peak with industrialization and urbanization (pp. 1). Whether this is correct or to what extend it is correct may need more data and references, but it is certain that prostitution has become more diversified, concealed, and complex with the changing nature of commercial sex markets [17,10,3]. The reasons why women enter the sex business and their lived experience in it thus becomes more complicated.
Liu has identified six different paths of women entering into prostitution, namely, influence by boyfriends or male partners, family or personal events, persuasion by relatives, influence by friends or co-workers, self initiation, and force/deceit/coercion (pp. 75). From her research she indentifies six out of forty women as victims of human trafficking (pp. ix). The involvement of the others in the sex business is usually due to more than one of the factors listed above. It is, however, difficult to form a causal relation between these factors and the decision to enter into prostitution. The reasons can be personal and occasional; the context varies, and the relevance is different. For example, in some cases, we may find the time when some special events happen (quarrelling or breaking up with partner/husband/boyfriend, conflicts with parents, rape, etc.) to be crucial if we try to relate it to the woman's involvement in prostitution. It may happen right before her decision to become a prostitute, and so becomes the triggering factor. The influence of friends/relatives/co-workers can be significant, but should not be overemphasized as we could not tell exactly how it plays a role in the woman's decision-making process.
The social, cultural and economic conditions under which these women live, and how these conditions exert influence in their lives and produce internal desires may be a more important aspect that scholars should pursue [13,5,14,3]. The conclusion that prostitution was by no means a woman's first choice (pp. 107, 168, 170) can be legitimate if we put it into a wider context of global economic order, the restructuring of capitalist production across the globe [6], and the structural and gender inequality exacerbated in this process. However, it may need further scrutiny if it is constructed upon a more micro-and regional-specific background. Liu has discovered that among the women who voluntarily adopt prostitution, poor economic condition is not the only or direct reason. Many of them did not choose between starvation and prostitution (pp. 107). Having better economic conditions, exploring more economic opportunities, earning money in an easier and quicker way are more often the major driving force.
Economic reasons are believed to be the major motivation for women to enter the sex business. For the women themselves this is also a simple and defendable explanation to put forward. Female migrants suffer from employment-related discriminations such as work segregation, low pay, long working hours, bodily regulation, poor working conditions, no insurance, etc., as well as institutional discriminations caused by China's complex household registration system. Even when they leave their factory work or other legitimate work and enter the sex business this situation may not be fundamentally improved. They report unstable income, low dignity, little sense of responsibility, danger and stigma. Although many of the dreams of these women is to quit the job and run a small business at home, very few can achieve it. They earn a better income in comparison with their former legitimate jobs, but many cannot save much money due to their consumption habits developed with their prostitution experience [3]. So why do they still want to go into the sex business if this job cannot guarantee stable income and better conditions, especially for those who work in the lower hierarchy of the sex business, such as street working and home-working women? Considering the stigma attached to prostitution, the alienation they have to experience, the possible physical harm, etc., the price they have to pay seems to be too high. What do they want from this job? What can they actually gain? If economic motivation is the only reason for their migration and entering the sex business, we are not able to fully answer these questions. We need to probe into their own aspirations and perceptions and take other motivations into account. Only then can we understand their life choice.
Scholars have developed a language of desire to look at this issue, pointing out that sexual and affective desires are the same important analytical aspects as material desires but they are often neglected [16,14,3]. A woman's sexual and affective desires are subtle yet powerful motivations for them to migrate, enter the sex business, choose their current lifestyles, engage in intimate relationships, etc. Liu has also cited Agustin (2005) and Davidson (2001) in support of this (pp. 54) but she finally reaches another conclusion that economic incentive is still the decisive factor (pp. 55). On this basis, some women did choose to enter prostitution voluntarily, and some took it as a first choice, not only to pursue economic improvement, but more importantly, to fulfill some of their sexual and affective desires. This shows the complexity and diverse nature of why women go into prostitution that we should notice and emphasize.
In the chapter 'Prostitution and Human Trafficking: Underlying Reasons' Liu draws a more complicated picture by pointing out the 'distal and proximal factors, as well as factors at the aggregated and individual levels' (pp. 146) that influence women's involvement of prostitution. These aspects are developed through her theoretical framework of Rational Choice Perspective (RCP). She first analyzes how uneven development, imbalances in development policy and lack of social welfare programs for the working class, lack of legal protection, and gendered effects of economic development on individuals may have an impact on prostitution. In the section of 'social factors' she discusses traditional views of the woman's body, sex attitudes, gender inequality of education, change in attitude to money, and how these may lead to the proliferation of prostitution. I found the part dealing with 'situational factors' particularly interesting. Liu points out that weak social control is another reason for prostitution to be prevalent besides the influence of friends/co-workers, and omnipresence of advertising and entertainment establishments. With the breaking down of the danwei (work unit) system, the growth of the commercial housing estate and the interruption of communities, the formal and informal social controls based on traditional neighborhood networks are reduced. Urban neighborhoods become what we call a 'society of strangers' [9] and so the probability of being identified is greatly reduced. According to Liu, the perceived risks associated with certain 'deviant' (pp. 166) behavior such as prostitution and crime is thus minimized, which further pushes the women into the business. The idea that prostitution stigma and risks function as formal or informal social control deterrents to prevent women taking up prostitute jobs sounds uncomfortable, although it does happen to some degree. The basis of this ideology lies in the understanding of prostitution as deviance and ultimately an 'unsavory trade' (pp. 167), that only those who 'have a value system and a view of the human body far different from those held by society at large' would 'consider it an acceptable alternative means of support' (pp. 55). If we acknowledge the complicated nature of this phenomenon and want to find a more open and multi-dimensional way to understand it, we should first treat prostitution as simply a kind of job option, and the women involved 'persons' instead of any other identities [7].
Using the Rational Choice Perspective to analyze women's involvement in prostitution is helpful since it looks at the issue from different perspectives ranging from historical to current, and from social to individual. This all-inclusive framework may help build up a more complete picture. Because of its 'soft' nature compared to traditional criminology (pp. 57), its emphasis on the 'situational', 'current' and 'immediate' circumstances, and its attention to the individual's perception of these circumstances, RCP is able to catch the subtleness of the decision-making process of prostitute women that is often overlooked. The problem is that RCP is usually used in criminal studies, explaining criminal behavior from a more humanistic perspective with interconnecting factors, telling how individuals make choices and decisions, and how they exhibit rationality during the process (pp. 56). This again, is putting prostitution into a weird position as either an offense or a crime, but in the book it is actually treated neither as a crime nor a normative behavior, and sometimes it is seen both as deviance and as an abnormal choice. The women involved are thus sometimes considered as 'offenders' if using the RCP framework, or as those 'fit[ting] the stereotype of women in the sex industry' (pp. 71). This may not have been the author's original intension to view them this way; this framework as well as some accounts and analysis of the author, however, may jeopardize the readers' understanding of her scholarly position.
The Rational Choice Theory adopts a belief that human beings are rational actors who weigh the costs and gains, means and ends, possibilities and limits before they make decisions and choices. But scholars have found in their studies on human agency that besides planning and strategizing with available resources, cognitive agency such as imagining, dreaming and desire actually play an essential yet 'quiet' and 'intangible' role in one's decision-making process [8,5]. This dimension brings out the mostneglected element in exploring prostitute women's lived experience in a modernizing and globalizing context: the desires that drive the women out from the rural villages and into the sex industry. From prostitute women's cases we can see that not all decisions are based upon rational thinking and weighing. Although most of the women are quite well-informed of the industry, they may enter with imagination, impulses, desires, as well as vague dreams, subtle longings, future aspirations, and sometimes even willful thoughts. With this in mind we may be more able to make sense of the choices and decisions that seem to be irrational, impulsive or difficult to understand. We thus need to develop the logic of emotion to supplement the economic logic in order to understand their life choices. We need to catch the 'unconscious', 'irrational' and 'imaginative' aspects of a woman's desires, which reflect subtle relations between individual longings and the social context that might be ignored in other frameworks.
The title suggests that the book is about migration, prostitution and human trafficking, which are the three main themes readers want to focus on. It perfectly spells out the internal logic of the book: the regional gap and imbalance and the inter-city migration is the important context of Chinese prostitution, and among the prostitute population, trafficking constitutes a severe problem that needs to be addressed. The book layout puts more attention on prostitution than the other two issues. The migration part of the story and China's prohibition policy against prostitution should be addressed at an earlier stage, and in a more coherent and complete manner for a better understanding by readers unfamiliar with China. The RCP framework contributes to our understanding of why women enter the sex business but falls short of explaining the trafficking side of the story. Trafficking in prostitution is difficult to define, since it may take on different forms and can happen at any stage of the job. As Liu said, it is a 'perpetual dilemma' (pp. 182) that is both hidden and controversial. With the lack of official data and complete statistics, the readers may want to know more about prostitution and sex trafficking, and this is supposed to be the main contribution and strength of the book, as stated in chapter 1 (pp. 25). However, as this present version shows, the book has not quite answered some of the questions raised at the outset, especially those concerned with sexual trafficking. For example, it would be worthwhile to know how many women are victims of sexual trafficking (pp. 13), the nature and extent of sexual trafficking in China (pp. 14), and what has been said and done in academic discussions as well as by policy makers. More discussion and analysis on this theme would be helpful, as well as a clearer picture of the current situation in China, especially the trafficking of women or children into prostitution, facts and numbers, women's experience in it, and actions against it. The author may also want to review additional works on prostitution and sexual trafficking in particular.