Human Trafficking and Labour Exploitation in the Casual Construction Industry: An Analysis of Three Major Investigations in the UK Involving Irish Traveller Offending Groups

Human trafficking and modern slavery are routinely framed as key threats facing society. Despite increased media, policy, and practitioner attention the evidence base remains underdeveloped. The numerous knowledge gaps include a lack of empirical studies and research into labour trafficking. Since labour trafficking is a complex and varied phenomenon, we chose to explore one subset of it in a systematic, detailed, and empirical fashion. Our focus was exploitation in the casual construction industry in the UK perpetrated by Irish Traveller offending groups. We used hard-to-reach data from three major police investigations, namely operational case files and interviews with senior investigating officers. Taking a qualitative approach, we disentangled behaviours and dynamics across three major stages in the trafficking process: recruitment, control, and exploitation. We identified key challenges associated with investigating this crime. In discussing our study’s implications for research, policy, and practice, we also include officers’ recommendations for future cases.


Background
Human trafficking, sometimes referred to as modern slavery, 1 has been associated with adverse effects on the social, physical, and economic welfare of individual victims, whole industries, and communities and even the security of nations (Kelly, 2002;Belser, 2005;Zimmerman et al., 2006;Home Office, 2012).
The trafficking literature remains, however, underdeveloped and has been heavily criticized for a lack of methodological transparency, scientific objectivity, and validity of research design and interpretation of results (Kelly, 2005;Laczko, 2005;Laczko and Gozdziak, 2005;Goodey, 2008). Additionally, the traditional focus on international movements and sexual exploitation has left both internal (domestic) trafficking and other exploitation types largely overlooked (Andrees and van der Linden, 2005;Laczko and Gozdziak, 2005;Winterdyk and Reichel, 2010).
Labour trafficking is now recognized as a priority both in the UK and internationally (Home Office, 2011bEuropean Commission, 2012;US Department of State, 2014). In the UK, the year 2014 saw not only a general increase in trafficking referrals but a particularly steep growth in suspected labour trafficking cases (National Crime Agency, 2015). In their ongoing research, Cockbain and Bowers (2015) found the most common exploitation type among the 2,727 trafficking victims 2 officially recognized in the UK from 2009 to 2014 was labour (44%): ahead of both sex (41%) and domestic servitude (12%). Concern exists around the lack of robust research evidence on labour trafficking in the UK (Home Office, 2007, 2011aCockbain et al., 2014).
'Labour' trafficking is, however, a broad category that encompasses diverse licit and illicit sectors. In the UK alone, known industries of exploitation include agriculture, food processing factories, construction, hospitality, forced begging, shoplifting, and benefit fraud. Given the clear legislative, regulatory, procedural and practical variation between such industries, responses are likely to require sensitivity to the specific sector(s) in question.
In this study, we pursued a narrow but in-depth exploration of a particular subset of labour trafficking: trafficking in the UK casual construction sector involving Romany, Gypsy, or Traveller (RGT) offending groups. Of course, not all traffickers in this industry are RGT and individual offenders' criminality should not implicate RGT communities as a whole.
Several factors had together highlighted the need to improve understanding of and responses to this particular form of trafficking. First, a new law designed to tackle labour trafficking had recently come into force: the Section 71 offence of 'slavery, servitude and forced labour' of the Coroners and Justice Act 2009 3 (since superseded by the Modern Slavery Act 2015, which contains an equivalent offence). The first convictions were achieved in 2012 in a case involving an Irish Traveller family who had exploited their victims under horrendous conditions in the casual construction industry (Topping, 2012). Similar investigations were underway.
Secondly, there was international concern around this phenomenon. Swedish authorities had repeatedly identified a distinct threat involving RGT-specifically, Irish Traveller-groups trafficking vulnerable adult males from the UK into Sweden and Norway for exploitation in tarmacking and paving (National Criminal Police, 2007;National Police Board, 2009. A joint project to tackle this problem was launched in 2008, involving EUROPOL, the UK, Ireland, Norway, and Sweden (National Police Board, 2009).
Thirdly, the UK Human Trafficking Centre (UKHTC)-part of the National Crime Agency and the UK's lead on counter-trafficking-had noted a rise in cases of trafficking by RGT groups in the casual construction sector. When developing this project, UKHTC staff in roles as diverse as case management, victim support, intelligence, operational support, and strategic oversight all expressed the opinion that this form of trafficking was on the rise, poorly understood and involved distinct challenges. Among these perceived challenges were issues linked to RGT communities' mobility and segregation. Hostility and discrimination faced by RGT groups from the population at large and poor relationships with the police in particular are both well-documented problems (Niner, 2003;Cemlyn, 2008;Kabachnik, 2009).
Very little has been published on RGT working practices, probably at least partially due to difficulties accessing research data and participants. From the little information available, we learnt of a purportedly long-established practice of recruiting homeless men (known as 'dossers') and exchanging food and board for their labour (Clark, 2002;Judiciary of England and Wales, 2012;Walsh, 2010). The criminal activity addressed in this article may plausibly lie at the upper end of a spectrum running from mutually beneficial exchange to extreme exploitation. Unsurprisingly, no figures exist on how common it is to keep 'dossers' or to subject them to extreme exploitation.
Trafficking in general is better understood as a process crime than a one-off criminal event (Aronowitz, 2001;Brayley and Cockbain, 2014). Consequently, one of our key aims was to disentangle common processes and dynamics involved in this form of labour trafficking across three fundamental stages of trafficking: recruitment, control, and exploitation. Our other aim was to identify key challenges associated with investigation.
Our exploratory study addresses four research questions, each deliberately framed in broad and inclusive terms: 1. How were victims recruited?
2. Through what mechanisms were victims controlled?
3. What forms did the exploitation take?
4. What were perceived as the key challenges in these investigations?
We used the standard definition of trafficking 4 set out by the United Nations (2000), which underpins counter-trafficking law and policy in the UK and many other countries.

Method
In this section, we introduce our sample, data, and analytical procedure.

Sample
There was no central repository of human trafficking investigations through which to identify relevant cases easily and comprehensively. Instead, we started by asking key UKHTC staff to identify as many cases as possible that met our inclusion criteria, namely investigations that: 1. addressed trafficking labour exploitation in the UK casual 5 construction industry; 2. focused on a RGT offending group; and 3. had a trial date set by the time our data collection began (November 2012).
UKHTC staff were well positioned to do so given their roles managing the national intelligence picture on trafficking, providing operational support to police forces and assessing potential victims (through the 'National Referral Mechanism'). As a safeguard, we also asked the investigative leads on each case identified about any similar cases of which they knew. Each independently named the others' cases, but no more, suggesting we had reached saturation. Our sample is nevertheless best characterized as a convenience sample since we cannot be certain that it covers all possible cases and that no biases affected case identification. In total, three investigations met our inclusion criteria: Operations Netwing (Bedfordshire Constabulary), Helm (Hampshire Constabulary), and Tundra (Gloucestershire Constabulary). All were pursued under the Section 71 offence. The Senior Investigating Officer (SIO) on each case agreed to participate in our research.
There were 15 defendants (hereafter the 'offenders') and 19 complainants (hereafter the 'victims') 6 overall, distributed as follows: Operation Netwing: seven offenders and eight victims Operation Helm: three offenders and five victims Operation Tundra: five offenders and six victims While we sought to identify as many possible cases for inclusion, our aim was not a primarily quantitative analysis but rather an in-depth, bottom-up, qualitative exploration of this particular type of labour trafficking. Accordingly, our priority was using rich empirical data to disentangle the complexities of this issue. While the sample size is small, our study is original and informative and makes use of notoriously hard-to-access data.

Synopses of the cases
Here, we provide synopses of the operations, including their initiation, investigative approach, and eventual prosecutions.
Operation Netwing. Operation Netwing was the largest and longest lasting of the three investigations and resulted in the first Section 71 convictions. Bedfordshire Constabulary launched Operation Netwing in Spring 2011, after a man alleged that an Irish Traveller family had promised him paid labour only to enslave, assault, and threaten him. He claimed other victims were still being exploited.
A 6-month intelligence-gathering exercise followed, including extensive covert investigation. During a coordinated strike day in September 2011, five suspects were arrested and 24 potential victims were removed from the site. Two additional arrest days ensued.
Six men and one woman were tried together in Spring 2012, primarily on charges of conspiracy to hold people in servitude and require them to perform forced labour (plus some assault charges). Of the seven, four were convicted, one was found not guilty and the jury was hung on the final two (a retrial was planned but later dropped).
Operation Tundra. Operation Tundra was the earliest of the three operations. Gloucestershire Constabulary began the investigation in Spring 2010, after several individuals came forward alleging labour exploitation at the hands of a local Irish Traveller family and the body of another suspected former victim was discovered. 7 The alleged exploitation predated the entry into force of the Section 71 offence. There was, however, intelligence suggesting other victims were still being exploited and a covert strategy was pursued to amass evidence of prosecutable offences. A coordinated strike day took place in March 2011, involving simultaneous arrests at eight locations across the UK. Five suspects were arrested and 19 potential victims removed from the sites.
Four men and one woman were tried together in Autumn 2012. All were convicted of conspiracy to hold people in servitude and conspiracy to require them to perform forced labour.
Operation Helm. Hampshire Constabulary launched Operation Helm in June 2011, after a man sought help from his local police station to return home to Romania. He claimed that heand others-had been held by an Irish Traveller family and forced to work long hours for no pay.
Since the suspects had no fixed domain and travelled frequently around the country, investigators favoured a short-term operation aimed at rapid disruption and enforcement. The strike day took place just 10 days after the initial contact with the police. Three suspects were arrested and 13 potential victims removed from the site. A fourth suspect was later arrested.
One man was tried alone for a public order offence and two others faced a joint trial on charges of requiring people to perform forced or compulsory labour. The former was acquitted and the latter were eventually both convicted in early 2013 following a guilty plea in their retrial.

Data
We used two main data sources for each investigation: restricted police operational data and interviews with the SIO. The diverse operational data included case summaries, demographic information about the victims and offenders, information on the offenders' prior criminal histories, and miscellaneous additional documentation. The victims' accounts proved a particularly rich source about the trafficking process. While the investigative case files contained various corroborating evidence, much remains unsubstantiated self-report data. This is a limitation of our research, albeit one that is very common to research into sensitive crime issues.
With each SIO, we conducted one face-to-face semi-structured interview, covering: the background to the investigation, including why it was instigated; a description of the investigative process and approach; key challenges encountered in the investigation; approaches used to overcome these barriers; things that might be done differently in future; and any other information the participant wished to note.
At the time of our research, the investigations were still 'live' in that the prosecutions were not yet complete. We were unable to record the interviews-generally best practice in interview-based research-due to the cases' active status, sensitive nature, and risk of prejudicing prosecutions. To ensure maximum fidelity in the absence of verbatim transcripts, we each independently took extensive notes during the interviews. We then combined our notes (which showed a high level of agreement) to produce a final set of interview notes. As an additional safeguard, we sent each force a draft of this paper to check for accuracy.
We drew on a wider set of observations and discussions to inform our research design and contextualize our findings. These included court visits to observe the prosecution of one of the cases and indepth discussions with a UKHTC analyst, an intelligence officer and a tactical adviser (all familiar with such cases), a senior representative of The Salvation Army (the agency that manages support services for trafficking victims in the UK) and a prominent academic with expertise around RGT communities.

Procedure
The theoretical framework for our study was situational analysis (Clarke, 1983;Wortley, 2012). In practical terms, the situational approach meant we: focused on the immediate context to the crimes; engaged primarily with the 'what', 'where', 'how', 'when', and 'who' rather than the 'why'; and explicitly considered the implications for crime reduction.
Our analytical method was content analysis since it is well established, broadly applicable, and well suited to systematizing the analysis of documentary files (Bryman, 2015). We drew up the explicit coding framework shown in Table 1 to ensure a transparent and systematic approach (Krippendorff, 2012). In addition, we drew out key information on the victims and offenders (e.g. age, gender, ethnicity).
Our primary concern was identifying broad thematic issues in the trafficking process and responses to it. The different data sources we triangulated did not always lend themselves to the provision of precise numerical results. 8 While we quantify findings where possible and appropriate, the study is above all a qualitative exploration of a complex issue.

Results
First, we outline key features of the offender and victim groups. We then present the results for each of the four research questions.
Thirteen of the offenders were male and two were female. Although an inclusion criterion was that the offending group(s) was RGT (a relatively broad designation), the offenders were all Irish Travellers (a narrower subset of RGT). Offenders were typically based in England in Autumn and Winter and travelled to varying degrees around the UK and abroad in Spring and Summer. Each case involved multiple offenders-groups of three, five, and seven respectively-linked by close familial ties of kinship and marriage. Offenders were aged 15-59 years old at the point of arrest, with a median age of 27 years. Intelligence suggested many of the offenders had long been involved in this form of labour trafficking, some starting as early as their mid-teens.
Ten of the 15 offenders had prior criminal convictions and together they had previously been convicted of 110 offences. Their criminal histories ranged from limited (one prior conviction) to extensive (27 convictions). A close examination of their criminal records pointed towards an opportunistic and generalist approach to offending, of which labour trafficking formed just one part. We identified only one prior conviction clearly linked to labour exploitation. 9 Prior convictions were most commonly for acquisitive offences such as theft from person, shoplifting, or burglary (n = 37) and motoring offences such as driving without a licence or insurance, with excess alcohol or while disqualified (n = 31). 10 All 19 victims were male and the majority (n = 16) were British. The others were Polish (n = 2) and Romanian (n = 1). All were white and none were RGT themselves. Victims were aged between 24 and 60 years old at the time of their release, with a median age of 47 years.
How were the victims recruited? At the point of recruitment, all 19 victims were unemployed and either living on the streets or in homeless shelters/hostels. Many were long-term homeless and very few were in regular contact with their families. None were registered as missing persons. Almost all the victims were heavily alcohol-dependent at the time of recruitment and three were drug-dependent (heroin).
Across the operations, recruitment locations were typically places rich in vulnerable adult targets: soup kitchens; homeless shelters/hostels; day centres; job centres; parks; and streets. There were some indications that offenders took precautions to minimize the chances of detection or intervention. They tended, for example, to wait outside soup kitchens rather than going inside, which may have helped avoid rousing employees' suspicion or being caught on CCTV.
The recruitment modus operandi was strikingly consistent both within and between operations. Lone targets or small groups were approached by one or more offenders and offered manual labour, almost always as part of an 'all-inclusive' package combining pay, accommodation, food, and tobacco. Some victims were also guaranteed regular alcohol, whereas others were promised help curbing their alcohol dependence. Pay was offered as a day rate, typically around £50 (with a range from £20 to £80).
The victims all went with the offenders willingly. According to their accounts, they were attracted by the promises of paid work, accommodation, and food. Some said they went intending only to stay for a few days or weeks, whereas others reporting having no particular preconceptions about how long they would stay. Once a victim(s) had agreed to the deal, he was normally taken directly to a traveller site.
Through what mechanisms were victims controlled? There was considerable variation between and within operations in terms of the period over which individual victims were exploited. Data on the approximate duration of exploitation were Issues raised by SIOs in the interviews Other possible challenges identified through analysis of the case files Human trafficking and labour exploitation Article Policing available for 13 of the 19 victims and ranged from 3 months to 15 years, with a median of 4 years. Just as no victims were recruited forcibly (e.g. kidnapping) none were subsequently confined physically. Instead, their continued compliance was secured through a mixture of violence, threats, psychological coercion, and emotional manipulation. Across the three operations, we identified seven common control mechanisms from victims' accounts and other supporting evidence. These mechanisms were used in varying combinations to maintain power over victims and can be broadly divided into negative, positive, and value-neutral techniques of manipulation. They are set out in Figure. 1 and briefly described in turn.
Threatened or actual violence. Physical assaults documented in the data included punches, kicks, slaps, scratches, and beatings with nearby objects like shovels or brooms. Some assaults were carried out on an apparent whim, whereas others were framed as punishment for perceived misdemeanours (e.g. clumsiness or slowness). Although most of the offenders were implicated in threatening behaviour, physical assaults were generally the male offenders' domain. Some were particularly brutal: one offender had to be restrained by his peers after nearly beating a victim to death. Similarly, certain victims bore the brunt of the abuse: one was described by his peers as a 'human punchbag'. These particularly brutalized victims appear to have served as a warning for others of what could happen should they disobey or displease the offenders.
In addition to actual violence, threats of violence, and death were a commonplace and apparently effective way of securing compliance and deterring escape. Victims in each operation spoke of knowing, or knowing of, individuals who had escaped only to be located through the offenders' extended family networks and brutalized as punishment. Some victims reported that the offenders had taken their photographs, possibly to help locate them if they escaped. The violence and threats left many of the victims terrified of recriminations.
Insults and humiliation. Most victims reported experiencing severe emotional abuse. They were treated as inferior and utter subservience was expected. Typical complaints included living as in the 'Middle Ages' and being treated 'like a dog'. Constant verbal abuse and humiliation eroded some victims' already low self-esteem and made them dejected, passive, and compliant.
Examples of humiliating treatment include one victim being trapped in the car boot and forced to sing nursery rhymes to his abusers' children and another being pressured into having a sexually explicit tattoo. In two of the three operations, the offenders shaved victims' heads. Victims were told it was to make them look tidy but investigators thought alternative motives were more likely, such as humiliation and control, hygiene, and ease of identification.
Isolation from other victims. Victims were split into teams controlled by a specific offender. Fraternizing with other teams was discouraged or outrightly prohibited. Another common but less overt way of isolating victims from one another was by using certain victims as overseers. Characterized by SIOs as 'alpha victims', these victims assumed a higher-status role that involved supervising, manipulating, and even threatening and assaulting their peers. Alpha victims were encouraged to report on planned escape attempts and other insubordination, to lie to new recruits about forthcoming payment and to talk about recriminations for attempted escapes. Alpha victims were seemingly selected for their character and/or physical traits, not because they had been with the traffickers the longest.
Supply of alcohol and tobacco. Most victims suffered from alcoholism at the time of recruitment and a few had drug problems. During the trafficking, drugs were rarely, if ever, supplied to victims. In contrast, alcohol was widely used to reward victims and ensure compliance, although strict controls on the amount supplied avoided compromising their ability to work. Most victims reported a reduction in their alcohol consumption once under the offenders' control. Those who claimed they had routinely drank up to 9 litres of white cider per day pre-recruitment, for example, said they consumed a maximum of 2 or 3 litres per day while being exploited. Tobacco was provided too but generally in meagre rations.
Withholding earnings. Although victims were recruited on the promise of a daily wage (generally of around £50), most never received a regular income-let alone the amount originally offered.
With the exception of one victim, 11 the maximum any victim reported receiving was the occasional hand-out of £5 to £10.
Victims were typically told that their wages were being 'looked after' until they were ready to leave, either for safety's sake or to stop them squandering them on alcohol. This appears to have been a deliberate technique to delay the realization that they would not be paid as agreed and seemed to reassure victims in the short term. If they continued to ask for payment, excuses, threats, or violence ensued. There were many reports of further victims outside the study sample who ran away after a few weeks after realizing little or no payment would be forthcoming. Promotion of a sense of family and belonging Figure 1: Overview of key mechanisms used to control victims.
11 He said he received £20-£40 per day, but this was thought by SIOs to be a lie to protect the offenders.
Human trafficking and labour exploitation Article Policing driving licences, birth certificates, and benefit cards.
As with the withholding of wages, they were told that their documents were being 'looked after' for safekeeping.
Provision of security and structure. Many victims described the security and structure provided by the trafficking situation as a key reason for remaining. Notwithstanding considerable hardships, they saw benefits in the routine, guaranteed accommodation, and a regular (albeit restricted) supply of food, alcohol, and tobacco. The prospect of returning to their former lives as homeless, destitute alcoholics was often seen as daunting, especially for those who had been with the offenders for years.
Promotion of a sense of family and belonging. Many victims said they were told they were now 'part of the family' and were forced to call their abusers 'ma' and 'pa'-even if they were several decades older than them. In turn, offenders sometimes called victims 'son'. This familial myth is particularly unconvincing since no victims were RGT; members of the settled community ('gorjas') cannot become RGT and fraternization with gorjas is often frowned upon for travellers. Nonetheless, we were told that defence barristers later characterized these interactions as proof of their clients' generosity and concern for the complainants' welfare. It seems more likely that the behaviour was intended to help erase victims' prior identity, build attachment, and infantalize them.
What forms did the exploitation take? Exploitation in the casual construction industry was an inclusion criterion for the investigations and so it was unsurprising that this was the primary labour market in which victims were exploited.
Commissions were generally carried out on a cash-in-hand basis without proper paperwork and the clientele were mostly individual home-owners.
Where licensed at all, offenders were registered as sole traders. The typical types of work performed by victims were: Tarmacking; Block paving; Other repairs to properties (e.g. painting walls); and Door-to-door canvassing for trade.
Arranging and facilitating construction work was the sole purview of the male offenders. Their roles were to negotiate prices with clients the victims had successfully canvassed, to oversee jobs and, in some instances, to perform skilled or technical tasks. In other instances, such skilled tasks were contracted out to paid labourers or victims were responsible for the entire job, regardless of their lack of training or experience. Alongside their primary and profit-generating exploitation in the casual construction industry, victims were expected to carry out domestic tasks and repairs at the traveller sites where they were housed. These tasks were arranged by the female offenders and typically involved cleaning offenders' caravans and picking up rubbish. The gendered roles played by male and female offenders seemingly reflect a broader gender-based division of labour in the Irish Traveller community. Domestic chores were completed around victims' already long working hours. Finally, a few victims were forced or coerced into fighting one another in informal boxing matches, on which the offenders and their associates placed bets.
All 19 victims were subject to unfair and unsafe working and living conditions, summarized in Table 2. Illustrative of the level of exploitation associated with this form of labour trafficking, each victim experienced most if not all of these substandard conditions. Additionally, victims' identity documents were used in one operation to make fraudulent benefit claims (worth around £200 per eligible victim per fortnight).
In two of the three operations, the offenders exploited not only the trafficking victims but also 10 or 12 hours a day (e.g. 7am-7pm) as standard.
Commutes added up to 2 hours each way on top.

Few or no days off
Manual labour 6 days a week.
On Sundays, victims canvassed for trade and/or complete domestic chores.
Victims expected to work through routine illness and serious injuries.

Few or no rest breaks
Common to go full working day without a meal or break despite strenuous conditions.

Little or no payment
Victims underpaid or not paid at all.

Insufficient training
No formal training and little informal training provided.

Lack of equipment
Insufficient tools and protective gear (e.g. hard hats).
Even basic warm and waterproof clothing lacking despite working through poor weather conditions.

Requirement to commit criminal offences
Victims made to offend in connection with their work (e.g. stealing equipment or driving without insurance).
A few victims convicted for such offences.

Makeshift living arrangements
At best, victims housed in old caravans often without gas or electricity. At worst, arrangements included a tent, the back of a van and a converted horse box Marked contrast to offenders' clean, fairly luxurious, and well-equipped static homes/caravans.

Unsanitary conditions
Conditions cramped, dirty, and hazardous.
Victims crowded into living spaces with no privacy.

Lack of access to hygiene facilities
Victims forbidden to use offenders' toilet facilities and expected to urinate and defecate outdoors.
Unlike offenders, victims wore same clothes and went without showers for days-even months-on end.
All victims on one case contracted scabies.

Inadequate nutrition
Victims usually fed once a day, sometimes twice.
Meals of low-cost and nutrient-poor food (e.g. chips, bread) Sometimes given offenders' leftovers or out-of-date food.
Human trafficking and labour exploitation Article Policing paying clients. Examples of this third-party exploitation included: over-charging clients (especially those with learning difficulties or mental health issues); deliberately causing damage and then charging to fix it; committing distraction burglaries; and repeatedly targeting certain clients for superfluous and overpriced jobs. In the third operation, however, police said the offenders offered good quality labour at below market rates. Although a detailed economic analysis was beyond our study's remit, it is worth highlighting a few key financial points. To give a sense of the potential scale of income, consider a common job performed by the victims: tarmacking. According to victims' and paying clients' reports, standard charges for tarmacking an average-size driveway ranged from £1,000 to £3,000 (although the most vulnerable clients were charged as much as £10,000). Its completion generally took the equivalent of eight 'man days' (e.g. four victims working for 2 days), with victims working long days under rushed conditions. While fluctuations in market demand mean the flow of work was not necessarily steady, victims reported generally being busy. A very rough estimate based on this information is that a victim could bring in £125-£375 per day 12 when working, highlighting how lucrative this type of trafficking might be. Of course, this rough calculation does not account for inevitable outgoings (e.g. building materials, petrol, victims' upkeep). Compared to legitimate businesses, however, offenders would have saved on non-payment of victims' national insurance, pension, or other employer contributions. Additionally, virtually all transactions with clients were done on a cash-inhand basis, allowing for easy tax evasion.
Routine searches and financial investigations showed many of the offenders to be wealthy, despite typically claiming to be economically inactive. Commonly identified assets across the cases included: large sums of cash; considerable bank balances; vehicles (including luxury models); expensive caravans and static homes; houses; land (e.g. caravan sites); and other consumer luxuries (e.g. jewellery). At the time of our research, plans were underway for asset recovery. Confiscation orders were later issued on Operations Netwing and Tundra to total values of over £0.4 million 13 and £2.3 million, 14 respectively (BBC, 2014a,b;Crown Prosecution Service, 2015). It is unclear whether confiscation orders were sought on Operation Helm.
What were perceived as the key challenges in these investigations? Although all three operations led to successful convictions and 11 of the 15 offenders were convicted Most victims reported decline in quality and quantity of food.
Victims on one operation found to be clinically malnourished (but link to prior lifestyle could not be excluded).

Lack of access to medical care
Access to routine medical and dental care denied, even for those with prior mental and physical health issues.
Emergency medical treatment granted in extreme cases only. 12 The lower end is based on a £1,000 charge for a driveway, the upper on a £3,000 charge. 13 All four convicted offenders convicted were subject to confiscation orders. 14 Four of the five convicted offenders were subject to confiscation orders. In August 2015, the CPS reported that it had already recovered almost £700,000.
overall, the SIOs identified in their interviews numerous challenges they faced in investigating this unfamiliar and complex crime type. We can group the diverse issues thematically into four broad categories, set out in Fig. 2.
Managing geographical dispersion. In all three cases, the locations in which victims were recruited, harboured, and commercially exploited were widely dispersed across the UK. The offender groups sometimes fragmented, with different offenders living on different traveller sites. In two operations, there was also some evidence of trafficking out of the UK for similar construction work in Scandinavia.
The SIOs saw numerous complications in this geographical dispersion. Criminal activity straddling multiple police jurisdictions was perceived as a barrier to building a coherent intelligence picture and to running investigations, affecting fundamental decisions such as which force should lead a case. Offenders' transient lifestyles had implications for investigative strategies and the execution of warrants. To illustrate, surveillance plans on one operation were delayed by 6 months after the offenders unexpectedly dispersed across the UK and Northern Europe. Meanwhile, executing simultaneous arrests at eight different locations across the UK required extensive interforce cooperation.

Managing geographical dispersion
Managing witness credibility issues Securing and maintaining vicƟms' co-operaƟon Handling the cases' complexity and resource-intensity ProsecuƟon Figure 2: Key challenges for investigation and prosecution.
Human trafficking and labour exploitation Article Policing Securing and maintaining victim cooperation. Engaging with suspected victims was perceived as a fundamental challenge in all three cases. On the various strike days, a total of 47 suspected victims were identified and taken from the different traveller sites. Our victim sample of 19 represents just 40% of these suspected victims. Of the remaining 28, 15 cooperated with the police but were not complainants and 13 refused to engage. While some of these additional suspected victims may have been misidentified as such, the SIOs saw the attrition rate as high 15 and as illustrative of difficulties engaging with vulnerable individuals who do not self-identify as victims and/or are terrified of recriminations.
Ongoing challenges were highlighted around maintaining vulnerable victims' cooperation over the months from the strikes to the end of the trials. Victims' fear of retributions and logistical challenges linked to their lifestyles were seen to translate into unusually great difficulties in securing their attendance at court. Once there, victims were said to struggle with processing questions formed with complex vocabulary or syntax and with undergoing the traumatic process of cross-examination by multiple defence barristers. The SIOs suggested the traumatic effects of victims' maltreatment during the trafficking were compounded by their pre-existing vulnerabilities (homelessness, alcoholism, marginalization, etc.).
Victims' welfare issues and support needs were seen greatly to exceed what victim support officers normally handle. Tensions were reported in juggling conflicting demands of ensuring victims' well-being and guarding against anticipated claims of witness coaching, inducements, and racist police conspiracies.
Managing witness credibility issues. The SIOs drew out several distinct but overlapping issues around witness credibility. First, they characterized managing the legal process of disclosure as particularly demanding and resource-intensive in these operations because of the number and nature of victims involved. Extensive records had to be reviewed and there were concerns that evidence of victims' 'bad character' would be used to undermine their credibility.
Secondly, the SIOs stressed that the effects of psychological manipulation are poorly understood. Despite appalling living and working conditions and no physical constraints, many victims had remained for long periods, made no escape attempts, and not capitalized on prior opportunities to disclose abuse. The SIOs described their-and the prosecutors'-concerns about how juries would respond to this seemingly paradoxical situation. Concerns were raised too about how defence barristers might-and did-try to reframe such inaction as proof that no exploitation had occurred.
Thirdly, they highlighted issues arising from incomplete and/or conflicting accounts given by victims. They saw it as a particular problem that some victims originally denied all abuse and only gradually revealed what they had undergone, as their trust of and rapport with the officers grew. Additionally, victims were said to struggle to distinguish accurately and consistently between events spread across long periods. While such gaps and inconsistencies may be attributed to fear or trauma, there were concerns that defence barristers would recast them as proof of the victims' dishonesty and unreliability, as they indeed did.
Finally, difficulties were associated with the way victims could be seen-and sometimes saw themselves-to have derived benefits from the trafficking experience: routine; stability; accommodation; and dramatic reductions in alcohol consumption. Such benefits were said to have become the focus of defence barristers' efforts to frame their clients as kind-hearted saviour figures.
Handling the cases' complexity and resource intensity. All three SIOs characterized their operations as unusually complex and resource-intensive overall, whereby contributing factors included but were not limited to the issues already identified in sections 'Managing geographical dispersion', 'Securing and maintaining victim cooperation', and 'Managing witness credibility issues'.
Difficulties were highlighted in building coherent cases when dealing with criminal activity that spanned long periods, involved multiple victims and perpetrators, and covered diverse offences (e.g. assault, fraud, theft, false imprisonment). Historic offences were seen to present particular challenges, not only because of difficulties gathering evidence but because some allegations long predated the enactment of the Section 71 offence that was the mainstay of the cases.
All three investigations used covert evidencegathering strategies, reportedly complicated by the sheer number of victims, suspects, locations, and vehicles involved. Deploying surveillance at the traveller sites proved particularly hard, frustrating efforts to gather corroborative evidence of victims' treatment in the places where the worst abuses allegedly occurred. Consequently, the prosecutions were said to rely heavily on victims' testimonies.
Further challenges were associated with the strike days themselves, in particular ensuring effective but safe enforcement (arrests and evidence gathering) at RGT sites where numerous people lived and hostility towards the police was expected. In one operation, over 100 officers participated in the strike on one site alone. In addition to police, other agencies identified as central to the strike days included medical professionals, third-sector agencies and environmental health. Running a smooth multiagency operation was seen to pose logistical challenges and the high staffing needs had cost implications.
A final complexity associated with these cases was a perceived lack of clarity around what constituted 'coercion' and 'forced labour', key elements of the offences pursued in court. SIOs described initially finding Crown Prosecution Service (2010) legal guidance helpful in understanding the issues and guiding evidence collection. They referred in particular to the lists it contained of indicators of forced labour (e.g. hazardous working conditions) and of behaviours that normally constitute evidence of coercion (e.g. threatened or actual violence). They said they were disappointed to learn later that these checklists had no legal standing.

Discussion
In this study, we capitalized on unusual access to a rich data set on a hidden and under-researched issue. Using investigative case files and interviews with SIOs, we delved into the characteristics of labour trafficking in the UK casual construction sector involving Irish Traveller offender groups.
The results indicate that this form of trafficking is harmful, complex, and perceived to present substantial investigative challenges.
As with any small-scale exploratory study, our findings should not be overstated. The small sample size and non-random sampling method may limit their generalizability to all trafficking of this particular type and we would certainly not expect them to be representative of trafficking in general. Additionally, crime is not a static phenomenon and offenders may adapt their modus operandi over time (Ekblom, 1997).
These limitations notwithstanding, we found striking similarities across the three cases both in the trafficking process (recruitment, control, and exploitation) and the perceived challenges for investigation. These broad-based commonalities have important implications for research, policy, and practice.

Implications for research and policy
The traffickers appear to have focused their recruitment efforts on certain types of locations, rich in marginalized individuals. The victims' broad age range suggests the offenders were fairly indiscriminate to age: as long as someone could physically work he was a useful commodity. Trafficking homeless, alcohol-dependent men can be seen as a low-risk, high-reward strategy: such victims' expectations, incentives to escape, and likelihood of being reported missing could reasonably be expected to be lower than the average person's. This aspect of the offenders' behaviour fits with the rational choice perspective on crime (Clarke and Cornish, 1985;Cornish and Clarke, 1986), according to which offending is generally the result of a rational decision-making process.
Not only did a high proportion of the offenders have previous criminal convictions but their conviction histories indicated that they were criminal generalists rather than committed specialists. These results add to a small but growing body of evidence showing that even serious and organized criminals tend towards generalism, indicating that the crimes they commit are influenced by the different situations they encounter and the various offending opportunities contained within them (Smallbone and Wortley, 2000;Wortley and Smallbone, 2006;Cockbain and Wortley, 2015).
None of the victims were forcibly recruited, unsurprising since force is seen as an unusual pathway to recruitment for trafficking in general (Kelly, 2002;Belser, 2005;Aronowitz, 2009; Organization for Security and Co-operation in Europe, 2010). The widely reported use of psychological coercion supports previous contentions that psychological techniques are important in controlling trafficking victims (see, e.g. Hopper and Hidalgo, 2006). The manipulation of victims through a diverse and unpredictable combination of positive and negative techniques recalls the literature not just on human trafficking (Bales et al., 2004;Brayley et al., 2011;Hopper and Hidalgo, 2006) but on sexual grooming, too (Berliner and Conte, 1990). Some of the other control techniques documented in our data (e.g. the retention of victims' identification documents, excessive working hours) are also widely used as indicators of human trafficking (International Labour Office, 2009).
The role of so-called 'alpha victims' recalls studies on sex trafficking that have emphasized how past and present victims may recruit, groom, and control other victims (Monzini, 2005;Brayley et al., 2011;Cockbain, 2013;Mancuso, 2014). The continued invisibility of alpha victims in policy debates may pose risks to their own and others' welfare.
The extremely poor living and working conditions found here are reminiscent of the violence, abuse, and neglect documented in a case series of 35 labour trafficking victims in the UK (seven of whom were exploited in construction) (Turner-Moss et al., 2014). There, 81% of the sample reported one or more symptoms of poor physical health and 57% reported one or more symptoms of post-traumatic stress. While we did not explore the health impacts of trafficking, our findings about the victims' maltreatment add further weight to calls to improve post-trafficking service provisions.
Also important in terms of victims' welfare was our finding that some victims were prosecuted (and even incarcerated) for crimes they were compelled to commit during the trafficking. This goes against the non-punishment principle of European counter-trafficking commitments, which the UK authorities have previously been criticized for contravening (The Anti Trafficking Monitoring Group, 2013).
Many of the challenges associated with intervention in this form of labour trafficking were similar to those raised in interviews with investigators and prosecutors involved in cases of internal trafficking of British children for sexual exploitation (Cockbain, Forthcoming). Key examples include difficulties managing vulnerable victims and witness credibility issues. These parallels suggest there may be some fundamental issues associated with tackling trafficking that apply regardless of the specific subset. Barriers to detection and intervention may be particularly acute when traffickers operate within marginalized and isolated groups, such as the Irish Travelling community. Amid growing pressure on forces to tackle trafficking/modern slavery, much greater understanding is needed of the costs and complexities involved.

Implications for practice
Our results highlight the importance of coordinated efforts and improved information sharing both within the police (across forces) and with outside agencies. Regional Organised Crime Units 16 may be able to help with sharing information across force boundaries. While the police have an important role to play in tackling this crime, it is not something they can or should do alone. Existing initiatives, such as the Government Agency Intelligence Network (GAIN), may prove valuable in helping bring together partners who can support future investigations.
We would argue that there is a clear and largely unmet role for harm-reduction activity. To illustrate, efforts to raise awareness, protect potential victims and detect and disrupt suspects could usefully be targeted at the type of locations where recruitment concentrated (e.g. soup kitchens and homeless shelters). Interventions could be rolled out to all such locations or-better yet-focused on known hotspots identified through local problem profiling. Meanwhile, the offenders' criminal versatility suggests a possible role for targeted disruption focused on trading standard violations, motoring offences, or tax evasion.
Despite our enthusiasm for prevention, we recognize that reactive enforcement is crucial, too. Table 3 contains a detailed but non-exhaustive list of tangible measures that might usefully be considered in future investigations into this form of trafficking. We grouped the measures according to the particular challenge they address. All were suggestions from one or more of the SIOs interviewed: they have not been formally evaluated but they represent experiential knowledge from real cases.
Researching criminal activity linked to a specific ethnic group can be fraught with challenges (Siegel, 2016): the same can be said of designing and delivering counter-measures. Although the scale of labour trafficking in the UK casual construction industry involving RGT offending groups remains unclear, it clearly requires a response if encountered by the police or other agencies. Although interventions to tackle crime should always be ethically sound, well-substantiated, proportionate, and carefully executed, it is arguably especially important to respect these principles when targeting particular ethnic groups.
Counter-trafficking is a rapidly evolving field both in the UK and internationally. Among the important legislative and policy changes postdating our study are the expansion of the Gangmasters Licensing Authority's remit into sectors including construction, the new Modern Slavery Commissioner position, Modern Slavery Strategy (Home Office, 2014), and Modern Slavery Act 2015. It remains to be seen whether and how such changes will affect trafficking (this sub-type and others) and counter-trafficking activity.

Conclusion
Despite increased prioritization of labour trafficking in the UK, there is very little concrete research evidence to inform policy and practice. Our exploratory study represents a small but valuable contribution to the empirical literature. Using novel data and focusing on one specific form of labour trafficking, we were able to explore in detail the complexities of the phenomenon and its perceived challenges for enforcement. The commonalities that we have identified across three major investigations help expand the understanding of this problem and stimulate future approaches to intervention. 16 Law enforcement organizations staffed by police from forces within a set geographical area. Improved information-sharing between forces to build a coherent intelligence picture and avoid duplication.
Early cross-force collaboration to identify which is most appropriate and best placed to lead an investigation.
Identification of patterns in offenders' movements that could guide investigative strategies.
Cross-force collaboration in executing simultaneous strikes across the UK.
Liaison with oversees authorities in cases involving trafficking movements from the UK.

Securing and maintaining victims' cooperation
Use of the 'reception centre' model to provide a centralized, safe environment to process victims, conduct interviews and ensure they gain access to facilities and services (e.g. showers, contact with housing officers).
Building rapport before interviewing victims to increase the likelihood of early disclosure of offences suffered.
Having separate teams of investigative and victim support officers to meet victims' welfare needs without compromising the integrity of the investigation.
Consideration of interviewer training about human trafficking and its effects.
Application for special measures for victims testifying in court.
Consideration of how best to deploy victim liaison officers to support victims in the lead up to and during trial.
Awareness of risk of witness intimidation, etc. and implementation of appropriate protective measures.

Managing witness credibility issues
Use of covert methods to gather corroborative evidence and reduce reliance on witness testimonies.
Documentation of control methods used (e.g. retention of ID, emotional abuse, threats of violence).
Involvement of environmental health officers in strike days to substantiate when victims' dwellings are unsafe.
Use of forensic medical examinations to document health issues and support victims' accounts (e.g evidence of malnutrition or past injuries consistent with claims of assault).
Consideration of a dedicated officer to manage disclosure and detailed log keeping by all to help dispel claims of improper conduct.