Migrant protection regimes: Beyond advocacy and towards exit in Thailand

Abstract International migrants are subject to many types of violence, such as trafficking, detention, and forced labour. We need an improved understanding of what protects migrants from such violence. The concept of ‘migrant protection regimes’ draws our attention away from formal rights advocacy and to both the informal dimensions of protection and the way migrants help determine the quality of protection they receive. ‘Migrant protection regimes’ are sets of rules and practices regarding who ought to protect whom. These regimes include formal rights to protection in the law and informal relationships that protect migrants from lawful violence by the state. They may be changed by ‘power grabs’, when sovereign actors seek to monopolise protection relationships, but also by ‘exits’, when migrants refuse to accept the protection on offer. The study demonstrates the value of these concepts by using them to explain an unlikely case: a change in laws concerning migrant protection in an authoritarian state: Thailand. Drawing on rich qualitative sources, the article reveals how, after a human rights advocacy campaign had placed migrants’ protection in jeopardy, a mass migrant exodus compelled the country's junta to offer migrants protection on better terms.


Introduction
Migrants are subject to many different forms of violence. They flee violence in their home countries to seek refuge abroad. They evade violence from traffickers and border guards en route to their destinations. Finally, migrants face violence at the hands of employers, police officials, and armed mobs in their destination countries. While there is a shared acknowledgement that contemporary migrants face an inordinate amount of violence, we have a limited understanding of how migrants obtain protection from such violence. 1 This article seeks to address this problem by identifying new forces that might make national elites introduce policies that afford migrants better terms of protection. The article specifically moves away from a reliance on liberal rights advocacy groups to emphasise the potential of informal rules and migrant agency to make policymaking conform with migrants' protection preferences.
Most research on migrant protection focuses on efforts to establish rights to protection in international treaties, and domestic laws, and efforts to encourage international organisations and national governments to uphold and enforce these rights. 2 This 'formal' approach has a lot to say about protecting migrants from unlawful forms of violence like human trafficking, forced labour, and xenophobic attacks, but little to say about protecting migrants from lawful state violencein the form of border controls, arrests, detentions, and deportations. This is a problem, because securing protection from state-enacted violence is, for many migrants, their abidingif not centralobject of concern.
Other research helps overcome some of the blind spots in the formal approach, by demonstrating how powerful liberal 'advocacy coalitions' can reduce the incidence of lawful state violence against migrants. 3 Liberal advocacy coalitions, made up of human rights NGOs and organised business, have reduced governments' capacity to arrest, detain, and deport unauthorised migrants, by challenging or undermining the control-oriented policies and laws advocated by 'nationalist coalitions' (composed of nationalists, security groups, and organised labour). Yet, this literature has not identified how migrants obtain protection when liberal coalitions are weak or divided, and/or political elites are able to ignore their demands. 4 This is an important gap in our understanding, because large portions of the world's migrants move between countries in the Global South, where liberal advocacy coalitions are comparatively weak and authoritarian governments are in charge. 5 For these reasons, this article seeks to reconceptualise migrant protection in a way that looks beyond formal rights and advocacy coalitions. In order to do so, I draw inspiration from two sources. First, critical research on international migration governance suggests that informal migration regimes shape the way ordinary officials and non-state actors deploy violence in the name of immigration control. 6 While this literature has demonstrated that the rules and practices that make up these informal regimes determine protection outcomes in a variety of micro-contexts and everyday encounters; it has yet to determine whether informal regimes can operate at a more macroscale, to constrain sovereign actors like national policymaking elites. 7 Meanwhile, historical sociological and feminist research demonstrates that the behaviour of sovereign actors is socially constructed by protection 'rackets'. 8 Yet, this research has yet to fully explore how migrants might meaningfully influence the type of protection that such 'rackets' afford.
This article builds upon these literatures, by introducing the concept of 'migrant protection regimes': sets of rules and practices regarding who ought to protect whom from whom in what ways and at what cost. Migrant protection regimes combine formal protections offered to migrants in national and international law and the more informal guarantees offered by networks 2 See, for example, A. Geddes, Immigration and European Integration: Towards Fortress Europe? (Manchester: Manchester University Press, 2000), pp. xii, 196; J. C. Hathaway, The Law of Refugee Status (Toronto: Butterworths, 1991). of smugglers, officials, and employers. The article then identifies two forms of agency that might change protection regimes. We are relatively familiar with the first form of agency, 'power grabs': when providers of protection seek to 'monopolise' protection services. We have been less attentive to the potential of the second, 'exits': when migrants opt out of protection relationships. The article demonstrates that the formal and informal dimensions of migrant protection regimes limit the behaviour of sovereign actors and that migrant acts of exit can change these regimes' defining rules and practices.
This study demonstrates the explanatory power of these concepts through an in-depth study of a deviant case: a change in the terms of formal migrant protection in Thailand. Thailand is a major migrant receiving country that has been ruled by a military junta since 2014. International migrants in Thailand are subject to multiple forms of state and non-state violence, and have few means of expressing their preferences for protection in the public sphere. Yet, in mid-2017 Thailand passed two starkly contrasting laws that had direct implications for the use of lawful violence against migrants and migrant protection. On 23 June 2017, the junta introduced an emergency law -The Royal Decree on Foreign Worker Management B.E. 2560 (2017)that mandated high penalties for illegal work: fines for employers and prison sentences for workers. Over the next few days, many foreign workers were fired and deported and others fled to the neighbouring countries of Myanmar and Cambodia. Then, a week after the first emergency law was introduced, the Thai prime minister introduced another emergency law -NCPO Order 33/2560 (2017)which placed a moratorium on the penalties for illegal work in the first decree. The junta then introduced an amnesty programme that helped unauthorised migrant workers legalise their terms of work and stay. Many of the migrant workers who had been deported or had fled, then returned to work in Thailand, and began to apply for permits.
This pair of laws constitutes a surprise for an advocacy coalition approach. Why would a military juntawhich is relatively immune to lobbying by migrant rights advocatesseek to protect and empower unauthorised migrant workers at the very moment when its own punitive law on immigration had begun to result in the expulsion and flight of those same workers? Had the balance of power between liberal and nationalist advocacy coalitions shifted over night? I argue that rethinking this set of events as an example of change in Thailand's migrant protection regime can help us resolve this puzzle. In short, the Royal Decree disturbed the unwritten rules of the Thai protection regime, thereby encouraging a range of actors to respond in ways that forced the ruling junta to repeal exclusionary sections of the law and offer migrants protection on better terms. While this analysis does not offer a 'recipe' to be followed in efforts to promote migrant protection in other contexts, it seeks to identify the ingredients that go into changing migrant protection outcomes, setting the stage for accounts that go beyond mere advocacy for rights.
The article develops this argument in five steps. Section One explains what protection regimes are and identifies two forms of agency that might change them. Section Two identifies the case selection and data collection techniques deployed in this study. Section Three describes Thailand's migrant protection regime. Section Four uses the 'migrant protection regime' concept to explain Thailand's contrasting immigration laws of 2017. Section Five draws out the implications of the study for scholarship concerning migration, violence, and protection. or undermine a violent act), and remedy (acts to provide redress or compensate for a violent act that has been committed).
The capacity to prevent, defend, and remedy violence is unevenly distributed across society. So, some will inevitably rely on others for protection. It is therefore useful to distinguish between 'protection providers'who sell/ award protectionand 'protection seekers'who purchase and/or obtain protection. 10 Protection services may be purchased in cash, for example, by paying a smuggler to take you across the border, or in kind, by working overtime for an employer who doesn't report you to the authorities. However, the value of protection relationships may also be built on less quantifiable dimensions of a relationship like 'trust': that a provider will not expose or subject those seeking protection to violence and/or exploit a protection seeker's vulnerability to threats of violence. 11 Protection is a 'racket' in which protection providers tend to be collectively positioned to dictate the terms upon which they offer services. There are three reasons why this is the case. First, protection markets tend naturally towards oligopoly, in which a limited range of actors have the capacity to offer meaningful protection from violent acts. As a result, those who seek protection have a limited range of choices of provider. 12 Second, and relatedly, people seeking protection generally lack the capacity to effectively deploy violence, and struggle to acquire and deploy leverage in their relationships with providers of protection, who commonly possess violent means. 13 Third, providers of protection often have a perverse incentive to increase or maintain people's risk of harm, thereby increasing the value of the protection services that they have to offer. 14 There are two additional reasons why protection providers are particularly well positioned to dictate terms to international migrants seeking protection. First, international migrants are all uniquely exposed to the threat of legitimate violence. International migrants may be distinguished from national citizens by what Nicholas De Genova describes as the condition of 'deportability', wherein they may be lawfully arrested, detained and deported for the simple act of remaining where they are and continuing to do what they do. 15 This condition most obviously applies to those migrants who migrate to a place without official authorisation. Unauthorised migrants must not only seek protection from violence at the hands of criminals, employers, family members, and corrupt officials, they are also exposed to the threat of legal arrest, detention, and deportation. However, the condition of 'deportability' also impacts those migrants who have acquired permission to stay or work in a national jurisdiction temporarily. While this group may be able to call on official protection against unlawful violence without necessarily fearing deportation, they can be distinguished from permanent residents and citizens in that they need to ensure that they will be protected from deportation in the future, and must fulfill a range of additional obligations, including maintaining a livelihood and not committing crimes, to reduce their risk of exposure. 23 Given that policymakers can annul or cancel temporary permitsas the Trump administration has recently done to Temporarily Protected Persons and DACA beneficiariesthis potential vulnerability can also rapidly become an actual one.
It is important to stress that migrants' condition of 'deportability' does not merely mean that they have limited protection vis-à-vis arrest, detention, and deportation at the hands of a state official. An array of other forms of vulnerability to violence either stem directly from, or are made more acute by, migrants' vulnerability to official deportation processes. For example, migrants become vulnerable to victimisation by potentially duplicitous smugglers because they cannot make use of legal forms of cross-border transport; migrants' exposure to the threat of arrest and deportation makes them less likely to contest violent acts in the workplace; and migrants' vulnerability to deportation encourages officials to violently extort them for bribes. While migrants may be formally protected vis-à-vis these ancillary forms of violence by rights in the law, the condition of deportability often limits the likelihood that they will be able to meaningfully avail themselves of such rights. Hence, migrants' liability to lawful deportation constitutes the centrepiece of a field of violent relations to which migrants are exposed.
In addition to being peculiarly exposed to violence, international migrants have a generic disadvantage in negotiating their terms of protection. This is because migrants have a deficitwhen compared to citizensof what Albert Hirschman describes as 'voice': rights to vote, protest, litigate, and bargain collectively. 16 Most migrants do not form part of the formal electorate in the jurisdiction where they require protection. In some countries, migrants' disenfranchisement is partially compensated for by powerful limits on the use of state violence, and institutions and actors dedicated to guaranteeing these rights, 17 or by universal protections on the freedom of speech, assembly, and collective bargaining. 18 However, the efficacy of courts, public protectors, and other institutions and actors capable of providing such guarantees, varies widely from one jurisdiction to another. In most contexts, migrants who are not authorised to reside and work in a country have few effective meanssuch as collective bargainingof giving voice to their individual or collective preferences or demands. As a result, migrants' terms of protection tend to be worked out among protection providers; keeping migrants' interests and likely responses in mind, but not necessarily listening to migrants about what terms of protection they prefer.
The assumption that migrants have significantly weaker rights both to protection and voice when compared to citizens, does not automatically lead to the conclusion that migrant protection will be equally bad in all places. Migrant protection may take multiple formssome of which may suit migrants' preferences more than others. In order to capture this variation, I introduce the term 'protection regimes', a concept that seeks to define the various ways in which acts of prevention, defence, and remedy are socially organised. A 'protection regime' is a set of rules and practices regarding who protects whom, in what ways, and at what cost. Constructivist approaches to the study of regimes in IR have long been attentive to the manner in which more formal, legally defined regimes are built upon a history of more informal customary practices. 19 Research on migration governance has argued that in the absence of a formal treaty, unwritten and unspoken exclusionary and exploitative rules, define how the agents of sovereign governments and private actors control migrant mobility and labour. 20 My understanding of 'regimes' extends this 'informal' approach further, to suggest that in certain contexts informal and formally guaranteed expectations of protection ought to be both understood as basic components of protection regimes as a whole. 21 Protection regimes vary in concentration: between monopoly and oligopoly. In monopolies a single actorusually a political elitedecides who is afforded protection and on what terms within a given territorial jurisdiction. However, most migrant protection regimes tend towards oligopoly, in which political elites share responsibilities with both (a) a variety of other official decision-makerssuch as police, immigration officials, and private subcontractorswho have legal discretion to determine protection outcomes; and/or (b) largely autonomous actors like humanitarian workers, smugglers, employers of unauthorised migrants and corrupt officials, who offer informal protection from legitimate violence. 22 While this tells us what protection regimes are and how they vary, in order to understand how protection regimes are formed and evolve, we need to pay attention to the agency of protection providers and seekers. Both providers and seekers of protection have reasons to change protection regimes in their favour. At a minimum, we can expect providers of protection to maximise their profits from the provision of protection, maintain or increase their market share, and seek relationships with actors they trust. We can expect seekers of protection to choose protection that is effective and inexpensive, and to secure relationships with actors that they trust. These expectations can be help us understand the nature and effects of at least two types of agency.
The first relates primarily to the distribution of responsibilities for protection. This involves actors making a 'power grab' by monopolising an oligopolistic regime. 'Power grabs' refer back to a process defined by Weber as the 'monopolization of legitimate violence', 23 and extended to the study of migration by John Torpey's work on the 'monopolization of legitimate movement': the long term struggle by states to arrogate to themselves the exclusive right to determineamong other thingswho may be forcibly removed from a given territory. 24 While monopolisation is commonly viewed as a completed process, it is better regarded as an ongoing struggle to distribute responsibilities and roles vis-à-vis the provision of migrant protection. In this vein, work on the 'migration industry' has observed the consistent drive by combinations of private, substate, and informal actors to procure and autonomously exercise the ostensible prerogatives of the state vis-à-vis violence and mobility. 25 This research suggests that policy elites can never simply assume monopoly, but must repeatedly reclaim sovereign prerogatives from a diverse range of actors.
Power grabs may, in turn, impact other providers' capacity to make a profit, impact their market share, and shape their relationships of trustgenerating pressure for them to alter their protection services and/or relationships. Power grabs may also impact seekers of protectiongenerating pressure for them to minimise their protection expenses, and potentially jeopardising the informal protection relationships that they value.
The second form of agency relates to the acceptance of offers of protection. Seekers of protection may create demand for new or reformed protective services through acts of 'exit'. My understanding of 'exit' draws directly on Albert Hirschman's formulation. 26 Exit consists of the decision to withdraw from a given relationship of power. This may involve physically 'voting with one's feet' and leaving the jurisdiction or area of operations of a protection provider, but may also involve severing a protection relationship or establishing a new one, while not moving at all. We know that in illiberal, non-democratic, and authoritarian contexts, those seeking protection routinely and en masse both (a) switch between informal and formal protection regimes; 27 and (b) flee places where they cannot obtain adequate protection. moving between protection regimes in different places 29 and between both informal and formal protection regimes in the same place. 30 When they exit, they may create demand for new services and/or pressure upon existing protection providers to renegotiate terms. 31 Both of these forms of agency are powerful but non-determinate. Power grabs and exits may create the pressure for changes in migrant protection regimes, but it is not clear what changes will result. For example, we don't know whether policymaker efforts to monopolise a protection regime will result in informal providers of protection inventing new ways of circumventing the law, or ceasing to offer protection services. We don't know whether acts of exit will result in providers of protection offering better terms of protection, or accepting their reduced capacity to extract value from protection services. The value of trust is a complicating factor here, resulting in actors' commonly making seemingly non-utilitarian decisions in order to stick with the 'devil they know'. Hence, understanding the outcomes of these forms of agency in specific cases will generally require interpretations that go beyond conventional rationalist forms of modelling, and that are attentive to the manner in which actors convey the intent behind their actions and interpret the meaning of others' behaviour.
To summarise, international migrants' expectations of protection are not adequately defined by national laws and international treaties: the places where IR theorists have commonly looked for evidence of international regimes. Migrants' expectations are likely to be shaped by a set of informal relationships and practices that partly constitute what I have called 'migrant protection regimes'. These regimes are unlikely to adequately cater to migrants' preferences, but are also prone to change. They do not only change in response to the lobbying efforts of advocacy coalitions for formal protection, but also in response to the concrete actions of protection providers' and seekers'. When those who protect migrants make power grabs and when migrants exit protection relationships, they both may generate pressures for change in the defining rules and practices of a protection regime.

Methods
I developed the concept of 'migrant protection regimes', through an exploratory study of a 'deviant' case. Deviant cases demonstrate a surprising value by reference to a specific theory. 32 In mid-2017 the Thai government passed two laws in a matter of days: The Royal Decree on Foreign Worker Management B.E. 2560 (2017) and the act rescinding the penalties in the decree: NCPO Order 33/2560 (2017). These acts and their surrounding administrative procedures, presented a stark contrast in the terms of formal protection offered to migrants. The Royal Decree included prison sentences for migrants working without authorisation in Thailand. The NCPO 29 B. Hoffman, 'Bringing Hirschmann back in: "Exit","voice", and "loyalty" in the politics of transnational migration', Latin Americanist, 54 (2010) Order placed a moratorium on these penalties and created provisions for the regularisation of unauthorised workers and legislative reform that would give these regularised workers greater freedom to change employers. I regarded this as a deviant case, because the dramatic shift towards formal terms of protection in migrants' favour ran contrary to the expectations of advocacy coalition theory. Advocacy coalition theory explains changes favoring migrants in formal protection laws as the outcome of the advocacy efforts of powerful liberal coalitions. Yet, the Thai military junta is relatively immune from lobbying of any sort, is deeply connected to a nationalist coalition, and migrant rights advocates have little if any influence or sway over its decisions.
In order to gauge whether an advocacy coalition approach might nonetheless help us understand this outcome, and explore the potential merits of 'protection regimes' as a competing explanation, I conducted fieldwork in Thailand from mid-2017 to mid-2018, primarily in the Bangkok Metropolitan Area. While there, I spoke to 58 key informants, including 22 government and international organisation officials, 6 employers, 3 brokers, 33 24 NGO workers, and 4 academics. Meanwhile, I hired a research assistant to conduct semi-structured interviews with 49 Myanmar migrant workers in Bangkok. 34 I then conducted document-based research including newspaper articles, government press releases, draft and final legislation, social media and used EU freedom of information procedures to gather a small selection of key resources on diplomatic relations between the EU and Thailand. Finally, I conducted observational research at NGO and policymaker workshops, government offices and business sites.
This combination of techniques had important limits. First, I was not able to secure interviews with officials in the office of the Prime Minister, Deputy Prime Minister, the former Labour Minister or the former Director General for Employment. I was able to partially compensate for this by interviewing senior officials in the Departments of Labour, Fisheries and the Navy who were directly involved in drafting and implementing the Royal Decree, and a range of nongovernment experts. However, I was not privy to parts of the decision-making processes that were ultimately responsible for the passage of the Royal Decree and NCPO Order. Second, I was not able to interview those migrants who had moved across the border in 2017. We asked all of the migrants we interviewed about their decisions during this period, studied responses on Facebook at the time, and conducted a follow up study of those who were leaving Thailand following the 2018 crackdown, but still needed to rely on secondary sources and inferences drawn from other data sources to account for their decisions. Third, I chose, due to ethical and security considerations, not to record interviews, but to instead take notes during conversations, and write extended summaries afterwards. This impacts on both the accuracy of the data and limits my ability to illustrate key empirical claims with direct quotations.
Despite these limitations, the data I have available lend substantial support to my argument that the concept of a migrant protection regime offers a more powerful explanation of the two laws than an advocacy coalition approach. The narrative below demonstrates how the empirical record supports my alternate interpretation of this surprising change in Thai migration laws.
Thailand's migrant protection regime: Oligopoly and the power of employers The discussion now begins to apply this approach to the case study by describing the set of rules and principles that defined the protection of migrants in Thailand prior to the passage of the 2017 Royal Decree. In brief: the protection regime in Thailand was oligopolistic, affording 33 'Brokers' are agents who facilitate migration and work processes in a wide variety of ways including assisting in applying for permits and documentation, organising transport, transferring funds, and mediating relationships between migrants and employers, smugglers, and officials. 34 In order to preserve the anonymity of respondents, the text references these interviews by their category (Official, Employer, Broker, NGO, Academic, and Migrant) and by assigning them a number that reflects the order in which the interviews were conducted. For example, the third NGO worker we spoke to is referenced as 'NGO no. 3'. employers considerable formal authority and informal discretion to decide which migrants obtained protection from a wide variety of actors that threatened them with violence.
Over the past three decades, several million migrants from neighbouring countriesand principally Myanmar, Cambodia and Laos, have migrated to Thailand in search of asylum, employment, and better livelihoods. 35 Most of these migrants lacked the right to claim asylum or access work permits and so many entered the country informally and found places to live and work without authorisation. 36 In doing so, they immediately exposed themselves to a range of actors with the power to inflict legal and illegal forms of violence. On entry, migrants faced the threat of arrest, temporary detention, and deportation. 37 Migrants also faced penalties of up to two years imprisonment for being in the country without a valid permit (s. 81, Immigration Act, B.E. 2522 (1979)). Every year, large numbers of these unauthorised migrants were arrested, detained, and deported from Thailand. 38 However, Thai police officials are reputed for petty corruption. Hence, migrants were often not prosecuted for immigration crimes prior to deportation, but instead commonly paid bribes to secure their release. 39 Migrants sought protection from arrest, detention, deportation, and extortion from other providers, including migrants, smugglers, officials, brokers, employers, and landlords. These actors assisted migrants to evade detection and acquire papers and helped to negotiate their release from custody. However, these informal protection relationships exposed migrants to further exploitation and abuse. Smugglers, brokers, and traffickers offered assistance in concealing migrants from state officials but also extorted migrants, detained them, and 'sold' them to employers in Thailand. 40 Employers protected migrants from officials by physically concealing them and paying bribes, but also took advantage of migrants' precarity, to make them work for low wages and under poor conditions. 41

Formalising employer protection
Policymakers have formalised elements of this protection regime. 42 Thailand signed memorandums of understanding (MOU) with neighbouring countries that provide for fully legalalbeit temporarymigration. The MOU permit (hereafter MOU) allowed migrants to legally cross the border and take up employment and residence in Thailand. Under the auspices of the MOUs, policymakers then created an array of temporary registration processes. Temporary Registration provides workers who have come to Thailand without papers and/or without official authorisation with the means to obtain passports or identity certification, temporary residence, and work permits. 43 (2008)). These increased penalties are particularly significant, because migrants possessing Temporary Registration permits have had few guarantees that their permits would remain valid, be renewed or replaced in a subsequent registration round.
These changes could be seen as a 'power grab', increasing the state's overall power to protect migrants. However, these moves are probably more accurately interpreted as merely perpetuating (while also reshaping) oligopoly, specifically inserting officials within the Ministry of Labour into a multi-actor system of protection. While this move gave Labour officials a new role in determining which unauthorised migrants would be protected against deportation, police officials retained the power to arrest migrants and added leverage to extort them for bribes.
More importantly, employers remainedin conjunction with police and brokers -key determinants of which migrants would receive protection. 44 Employers retained significant power to determine both whether migrants had the right to work, and whether unauthorised migrant workers would be arrested, imprisoned, and deported. Both the MOU and Temporary Registration permits made the right to work contingent on having an existing relationship with a specific employer and/or remaining with that employer. If a migrant worker left their place of employment, they risked invalidating their permits and exposure to arrest and deportation. Procedures to change an employer remained costly and complex and, in some instances, required the consent or participation of the original employer. Hence, migrants who were unhappy with their employment relationship, found it difficult to leave. 45 In addition to this direct ability to determine whether migrants were working legally, employers' had the power to determine whether migrants would be practically able to regularise. Acquiring the MOU and Temporary Registration permits was costly and complex. 46 These problems were exacerbated by the inability of the Department of Employment (DoE) to communicate rules effectively, the corruption of DoE officials, and the efforts of brokers to both inflate prices and exaggerate the complexity of application processes while routinely cheating their clients. 47 In this context, migrant workers relied heavily on their employers to pay and negotiate for their documents. 48 Employers tended to decide whether or not workers acquired legal documents or not, by not adequately informing migrants of their rights, withholding support, or actively preventing migrants from registering, sometimes by refusing to excuse them from work, so they could travel to the offices and complete the formalities. 49 Many employers then used this power as a form of leverage in the workplace. Some converted their payments for documents into debts that migrants had to repay and some limited workers' ability to leave or move freely by retaining possession of workers' permits. 50 Those employers who did not register their workers could offer a different form of protection: preventing their arrest by the police. Regularisation had not resulted in a slackening in police enforcement of immigration and labour laws. However, employers could: pay local police to either not inspect their businesses or not harass their employees in the immediate vicinity of the workplace; 51 acquire information from police informants about impending raids so they could warn their workers to hide; 52 pay bribes when migrant workers were arrested in the workplace; 53 and bail out migrant workers when they were arrested further afield. 54 Employers also threatened to call the police or immigration officials in order to exact concessions from undocumented migrant workers. 55 Migrant workers whose employers were unable or unwilling to provide protection, were commonly arrested and/or extorted. 56 This power was considerably enhanced by the fact that many migrant workers lived at their workplaces, and rarely moved out of their employers' realm of operations and local networks.

Migrant exit in the absence of voice
This did not mean that migrants have been forced to accept any protection arrangement on offer. This is due to the fact that migrant workers would commonly 'exit' protection relationships. Many would do so on entry to Thailand. The MOU permitting procedures were costly, slow, and tied workers to employers and fixed periods of stay. Hence, most migrants entering Thailand chose to bypass the MOU option and enter without authorisation with the informal protection of smugglers, and in the hope that they might be able to independently find an employer in Thailand who could help them to subsequently regularise their stay under a Temporary Registration scheme. 57 Opting out of the MOU process enhanced migrants' reliance on employer protection. However, migrants could also leave abusive employers, often en masse, in search of better terms of employment and places to stay, even though this move left them vulnerable to arrest. 58 Such acts impacted upon employer behaviour, rewarding those who offered better terms, and sanctioning those who didn'tin the form of the costs associated with recruitment and training new staff. 59 Other migrants opted to leave the country or abandon their old identity documents and permits in order to re-enter the country, or resurface at another office on different terms and as a new 'person'. 60 They also simply opted not to re-register themselves, because of the costs involved. 61 While some migrant workers were able to choose between protection services, they did not acquire the sort of 'rights' that would enable them to voice demands for improved terms of protection. 62 Some migrants sought the assistance of migrant NGOs and IGOs, who advocated for policy reforms on their behalf and assisted them to report and litigate abuses. On rare occasions of particularly egregious forms of abuse, they also sought outsometimes with NGO supportthe assistance of state officials and called upon its arbitration and relief processes. 63  and in Thailandthat government officials were more likely to exploit their vulnerability to harm than to offer them meaningful or lasting aid. 64 One crucial form of voice that migrants lacked was the power to collectively bargain. 65 Labour organisation in Thailand has long been weak, with Thailand reporting some of the lowest union participation rates in the world. 66 Successive Thai governments, whether democratic or military in composition, have all sought to keep labour disorganised. 67 While migrant workers were permitted to join unions, they were not allowed to form their own unions, or take leadership positions in existing unions (s. 88, Labour Relations Act, B.E. 2518(1975). Since the junta took power in 2014, the scope for organising labour has only narrowed further. 68 This is crucial, because it helps us understand the limited means available to convert migrants' individual capacity to exit into a more powerful form of collective leverage vis-à-vis decisions over their terms of protection.
There are two key points to take from this portrayal of the Thai protection regime. First, while Thai government agencies exchanged migrant protection responsibilities, employers' have retained considerable power to determine which migrant workers are shielded from arrest, detention, and deportation. Employers' role and power vis-à-vis migrant protection only becomes fully evident when we pay attention to both the formal and informal dimensions of the protection regime. Second, while migrants have limited ability to voice demands for better rights to protection, they routinelyalbeit silentlyexpress their support or opposition for the different protection services on offer, by exiting protection relationships.
These two points help us to understand why an advocacy coalition approach might run into trouble in Thailand. The various alliances among protection providers and seekers in Thailand tend to scramble conventional assumptions about alliances among a 'pro-migrant'/ liberal and an 'anti-migrant'/nationalist coalitions. Businesses sometimes align with migrant rights by protecting them vis-à-vis state officials and sometimes do this in conjunction with the 'nationalist' police officials. However, business and police officials can also turn around and together exert leverage against migrants in the workplace. As a result, while migrants often relied heavily on employerstheir 'liberal' coalition counterpartsthey also commonly work against their employers' interests by exiting the workplace in search of protection on better terms. In the next section, we will see how these nuances were lost on the EU officials seeking to promote migrant worker rights in Thailand, and how their poorly calculated interventions resulted in a major shake-up to Thailand's migrant protection regime.
The European Union provokes changes in the Thai protection regime From 2015-17, a migrant rights initiative challenged the core principles of the Thai protection regime. More specifically, human rights groups encouraged European Union (EU) officials to use the threat of trade sanctions to compel the military junta to take greater responsibility for protecting migrants from abuses in the workplace. Unlike the US State Department, that had long used the Trafficking in Persons Report to pressure Thailand for change, EU officials had no dedicated diplomatic tool for this purpose. Instead, EU officials developed a bespoke mechanism for promoting change in migrant rights policy under the auspices of its fishing laws, specifically the Illegal, Unregulated and Unreported (IUU) Fishing Regulation. 69 The IUU regulation is silent on migration and labour rights issues, 70  from countries found to be in persistent violation of European fishing regulations. Between 2013 and 2015 researchers and journalists had been revealing widespread abuses against the Thai fishing industry's mostly foreign-born workforce, including accusations of forced labour, trafficking, and slavery. 71 In this context, international human rights and labour rights NGOs successfully lobbied the European Commission to use the threat of trade sanctions to advocate for laws and law enforcement to protect victims of trafficking and forced labour, and the right to organise. 72 Some of the migrant rights NGOs that we spoke to believed that enhanced formal rights to protection from violence would mean little if migrants were unable to more actively motivate for their preferred protection outcomes. 73 However, they failed to convince EU officials to push for major improvements in the collective bargaining rights that migrants would need to do so. Instead, EU officials opted to pressure their Thai counterparts to introduce and implement laws to protect victims of trafficking and forced labour and supported new penalties for employers who hired illegal workers. The idea behind the second measure was that by cracking down on employers who hired illegal workers, the government might put an end to both unauthorised migration and work, cutting off the trafficking supply chain 'at its source'. This approach was first applied to the fishing sector in the Royal Decree on Fisheries B.E. 2558 (2015), which fined fishery businesses found to be illegally hiring foreign workers 400,000-800,000 baht (ss. 124,153).
Efforts to apply this approach to the fishing sector resulted in many deportations of migrant workers and the closure of a lot of Thai fishing businesses, butdue to the limited capacity/willingness of Thai officials to effectively pursue and prosecute human rights violations, and limited ability of migrants to make their own claims for protectiondid not result in any convictions for trafficking or forced labour. 74 Despite the poor outcome of this policy in one economic sector, EU internal documents reveal that officialsstill posing the threat of trade sanctionsthen advocated for the same approach that had been adopted in the fishing sector to be extended across the economy as a whole. 75 EU officials had not invented the idea of introducing a penalty on employers for hiring unauthorised migrant workers. Over the previous decade, Ministry of Labour officials had repeatedly attempted to eliminate the option of Temporary Registration system and to compel all remaining undocumented migrants to return to their home countries and re-enter under the more costly and cumbersome MOU scheme. 76 Under the military junta, this approach acquired the support of security lobbies, who viewed Temporary Registration as a magnet for informal migration and viewed the presence of large numbers of undocumented migrants as an inherent 71 security problem. 77 However, the local NGO and international organisation officials we spoke to concurred that large corporations in Thailand were extremely powerful and well connected to the current regime, 78 and pro-security lobbies had never been able to push anti-employer laws past Thai business.
It was the EU's threat of trade sanctions that settled the issueprompting the Prime Minister to use his unique ability to pass a Royal Decree to introduce a law that would pose threats to any actor that was part of an informal migrant protection relationship, including both employers and migrants alike. The main new provisions of the 2017 Royal Decree on foreign worker management were s. 119 and s. 122, which levied a fine of between 400,000 and 800,000 baht (10,000-20,000 euros) on employers for each foreign worker they hired without a valid permit and s. 101 and s. 102 which mandated a prison sentence of up to five years for anyone found working illegally. The provisions regarding prison sentences were not greater than the penalties in the 2008 law, but the fact that they were not adjusted downwards also reveals that EU officials had not aimed at protecting migrants from violence per se, but only its unlawful forms. This difference becomes crucial in the next section, where we learn how migrants interpreted this combination of changes and continuities in the law.
In summary, a migrant protection regime framework reveals the potential for the EU's formal rights advocacy to have ambivalent outcomes. While EU officials had compelled Thailand to conform with international standards of protecting migrants against traffickers, they had simultaneously goaded the government into an attempted 'power grab' that jeopardised migrants' primary protection relationship, while offering no enhanced protection vis-à-vis the primary violent threat most migrants encountered in their everyday lives: the police.
An advocacy coalition approach would struggle to detect these nuances because, again, the relevant actors do not neatly align into liberal and national coalitions. One component of the supposed 'liberal coalition' (anti-trafficking human rights advocates)teamed up with members of the supposed 'nationalist coalition' (the security lobby) behind a law that was contrary to the interests of another member of the 'liberal lobby' (business) and another member of the 'nationalist coalition' (organised labour). As we shall see, this not only leaves this approach without an explanation for the passage of the Royal Decree, but without the ability to interpret how protection providers and seekers responded differently to the law following its passage or why these responses resulted in the junta passing a secondmore migrant-friendly law.
A migrant exodus compels the junta to pass a migrant-friendly law The fine on employers was years in the making, but was smacked down in a week. The Royal Decree came into effect on 23 June 2017. By 30 June the prime minister had signalled his intention to revisit the provisions containing the fines on employers, formally issuing an NCPO order annulling the offending provisions on 4 July. The delayed mobilisation of Thailand's business lobby partly accounts for this turnaround. As soon as the government publicly announced the details of the Decree on 27 June, business groups responded negatively in press releases, reported comments in the media and through backchannels to senior government officials noting the potential for labour shortages and significant economic losses. 79 Yet, the sheer power of the business lobby does not account for the speed or nature of the new law. Business groups had more than a year, after the first version of the law was introduced in the fishing sector, to prepare to act. Draft versions of the law had been shown to business groups months in advance of its release and they had not mobilised to reverse the government's course. 80 One explanation for this inaction is that big businessesthe actors who exert disproportionate and more direct influence on the juntawere less likely to be dramatically affected by the law because they didn't make use of undocumented labour and/or could stomach the fines. 81 The power of the Thai business lobby also does not wholly explain the process of legislative change that took place and would follow. In addition to reviewing the fines against employers, the new NCPO Order also suspended five year prison sentences for illegal workers (Clause 1). Then, three days later, the Ministry of Labour initiated an expansive registration process, giving all migrants working in the country enhanced opportunities to regularise, and making it significantly easier to register for those workers who did not have employer support. 82 Finally, in the revised version of the emergency law, migrant workers holding these documents were given greater capacity to change their place of work, by simply notifying the Department of Employment (Royal Decree on Foreign Worker Management B.E. 2061 (2018) (ed. 2). In essence, the junta was not simply seeking to mollify corporate interests, but had produced a law that gave migrants the opportunity to substantially improve their protection arrangements, significantly reducing their dependence on officials, brokers, and also: their employers. What explains this attempt to make a more migrant (and not simply business) friendly law?
NGO's relative strength does not explain the outcome. These actors were important in helping to craft the government's new position. However, our interviews with NGO workers revealed that they have been pressuring Thai governments for decades to acknowledge the difficulties that migrants experience in becoming and remaining legal, and pushing for an approach that allows migrant workers to freely choose their employers without substantial success. 83 Their institutional leverage in Thailand is much smaller than that of business and has been further reduced under the junta. 84 They certainly do not have the institutional leverage to take on powerful corporations. Despite being shown previous versions of the decree, they had also been unable to motivate for changes to provisions affecting migrants. 85 Something happened between 23 June and 30 June that significantlyalbeit probably not permanentlyaltered the balance of forces in migrant workers' favour.
My approach suggests that in order to understand these outcomes we need to observe the way the junta's 'power grab' compelled ordinary actors to respond, and how these different responses culminated in the mass exit of migrant workers. First, the police took advantage of their new powers under the law. Police and immigration officials independently launched a series of raids on migrant workplaces and homes. In several parts of the country, police and immigration bureau officials conducted raids on workplaces and arrested and deported migrants without papers. 86 These raids were not coordinated by the agencies responsible for drafting and passing the emergency law. The prime minister had not instructed the national security agencies to lead such an action. Our interviews with Ministry of Labour officials suggest that their Ministry did not have any plans to significantly or suddenly transform enforcement procedures. 87 There was no interdepartmental coordination of enforcement activities or special resources set aside to ramp up enforcement measures. Our interviews with migrants suggest that raids did not occur in all parts of the country. 88 It is unclear whether these enforcement actions led directly  3;NGO no. 18;Migrant nos 14,15,22,23,24,32,39. 87 Official no.10. 88 Migrant nos 28, 31, 33, 36, 38, 41, 43. to an increased number of people being deported during this period. This is partly due to the fact that many Cambodian and some Myanmar migrants were reporting themselves to police and immigration officials in order to be transported back to their relevant border posts. 89 In short, the junta's passage of the Emergency Law was not signalling an intention to 'crackdown' on unauthorised work.
According to one prominent migrant rights activist, rather than a planned and coordinated crackdown, these raids and arrests may be better seen as opportunistic efforts on the part of local officials to capitalise on the additional leverage afforded to them by the new penalties in the lawand particularly the fines against employers. 90 The emergency law provided these officials with an opportunity toalbeit temporarilyexert a new form of pressure vis-à-vis employers, ramp up their immigration enforcement outcomes, and extract additional value in the form of bribes.
While policymakers had not planned a crackdown, the reporting of a small number of raids in the news and on social media helped to create the impression that the Royal Decree was going to be enforced in a more comprehensive manner. This helped shape the response of migrants' key source of protection: employers. Our interviews with migrants suggest that many employers encouraged their workers to remain, promised protection against any raids, informed their workers when raids would occur, and bribed officials in order to secure migrants' release. 91 In these respects, and in many places, the status quo in the Thai protection regime proved resilient throughout the course of the week. Employers assumed their traditional role of seeking the collusion of police officials to protect migrants from arrest, detention, and deportation.
However, other employers began firing migrant workers or instructing them to go home. 92 The high fines radically reduced the economic incentiveparticularly for medium and small enterprisesto protect migrants without permits, and made employers of unauthorised workers the targets of economic penalties for the first time. When combined with the sudden passage of the law, employers feared that they would be unable to come to an accommodation with police officials. It is plausible to suggest that employers also anticipated that non-local officials might lead the raids, reducing their ability to call on long-standing protection relationships with local Police and Labour officials, to secure their workers' release. According to the officials and NGO workers we spoke to, these considerations encouraged many employers to panic and shed themselves of the risk of prosecution through mass firings. 93 These two responses wereeven when taken togetherinsufficient to force the junta's hand, but their impacts were soon amplified by a third: migrant exit. Here, it is important to note that most migrants learn about changes in immigration policy and practice through word of mouth 94 or based on what they read on their social media feeds. 95 In this context, a small number of migrant advocates played a significant role in making flight appear to be the most suitable course of action. Some of these advocates worked for pro-migrant organisations while others were brokers. A few had become online celebrities of sorts, registering hundreds of thousands of hits for some of their posts. In the process, they had become the primary source of migrants' information on developments in ( After the passage of the Royal Decree, many of these activists' Facebook posts depicted a worst-case scenario. First, the posts mistook the law for an indication of the government's intention to more harshly penalise unauthorised workers. The Royal Decree mandated a prison sentence of up to five years for anyone found working illegally (ss. 101, 102). However, these provisions merely maintained the penalty at the same rate as the 2008 law. Yet, the activists presented the five-year prison sentence to migrants as if it were a new law and indicative of a tougher approach. 97 Second, the posts circulated fearsome images of the crackdownincluding videos of raidsto emphasise the threat that enhanced enforcement would present to migrants. 98 Third, the activists advised their colleagues without permits to either hide or return home to apply for a permit within the MOU system. 99 Two of these posts took the case one step further, suggesting that an exodus was the only way to achieve progress. By leaving en masse, the migrants would cause a massive disruption to the economy and force the government to change the law. 100 These posts were not prophetic, nor do they evince a sense of common political purpose or organisational capacity among migrant workers. However, in practice, they closely mirror the final impact that the ensuing exodus would have.
Once the reports of raids and firingsthe collapse of migrants' core protection relationshipswere amplified by social media, migrants began to exit. Migrants without permits began to leave their workplaces, places of residence, and jurisdictions en masse. Many paid brokers and drivers to smuggle them out to the border and out of the country. According to one official at an international organisation, significant numbers of the Cambodian migrants who left during this period combined these strategies, reporting themselves at local immigration detention centres and using deportation as a free ride to the border post. 101 Hence, while we see a substantial rise in deportation numbers during this period, much of this volume may be accounted for by these 'self-deportations'. 102 Others simply did not turn up for work, left their places of residence, or fled to other parts of the country. 103 It is impossible to develop an adequate numerical portrait of this exodus, but some of the data reflecting increases in movements across key border posts with Myanmar and Cambodia indicated that, within a few days of the announcement of the law, tens of thousands of migrants had departed and many others were 'in hiding' or simply not showing up at work. 104 This collective migrant exit helps us understand why the junta decided toin addition to reviewing the fines on businessalso revisit its policies on migrant registration and penalties for illegal workers. Here, it is important to note that the prime minister did not have to wait to see how these mass departures would impact both the economy and the government's credibility. In 2014, as one of its first acts in power, the military junta had conducted mass arrests and deportations of Cambodian workers. In that case, over 200,000 Cambodians left the country, bringing whole sectors of the economy to a halt, and disparaging the newly minted junta's international reputation. 105  that we spoke to, with this outcome etched in the junta's recent memory, top officials urgently wanted migrants to return to work, and were willing to make major concessions to ensure that they would. 106 This created the space for domestic migrant rights lobby groups to make headway in ensuing revisions of the law, demanding that migrants would not only need additional opportunities to register, but greater freedom to choose their employer and place of work. 107 The exodus did not create the awareness among government officials of the importance of foreign workers to the economy, but it appeared to have demonstrated the government's inability to compel them to regularise on terms dictated by officials and suitable to employers. 108 This account further demonstrates the value of paying attention to the informal dimensions of the Thai protection regime. The Royal Decree was not simply a change in the formal regime, but a move towards formalisation and monopolisation. By challenging a core informal understanding in its migrant protection regime: that employers decide who is protected and who is not; the junta was making a power grab. Power grabs may encourage other actors to respond, and this one encouraged: (a) usurpatious police officials to conduct raids; (b) risk-averse employers to fire their workers; and (c) wary migrants to flee. This approach also helps us understand why the junta quickly caved when the exodus began. Not only was the junta's top brassunlike its pro-security lobbiesonly taking on employers because of EU pressure, it was unwilling to bear the costs of a fight it had never wanted to pick, and instead opted for more generous migrant protectionsand further tolerance of informal migrationas a means of damage control. An advocacy coalition approach would struggle to understand or explain this sudden change. First, it would not recognise the significance of the differential gains of the competing members of the 'liberal' coalition, with migrants increasing their autonomy vis-à-vis their erstwhile protectors: employers. Second, it would not be able to explain why these changes suddenly took place over the course of a week, with little evidence that members of this 'coalition' had successfully advocated for the change.
These changes were, when we consider the scenario envisaged in the Royal Decree, a dramatic improvement for many migrants in terms of protection, providing them with greater freedom of movement in the country and the capacity to change employers. As a result, it is unsurprising that large numbers of migrants subscribed to the regularisation process. Many of our migrant respondents noted the advantages of becoming documented, in terms of greater leverage vis-à-vis employers, freedom from arrest, and freedom to move. At the same time, these improvements did not magically move migrants out of a general condition of deportability, or eradicate their reliance on employers, brokers, police officials, and smugglers to protect them from violence in Thailand. These everyday terms of protection continue to be worked out among formal and informal providers and seekers of protection through the deployment of power grabs, exits, and no doubtother forms of agency that I have not adequately encountered or fully described in this piece.

Conclusion
This study of a sudden and dramatic change in Thai migrant protection policy reveals how paying attention to the informal side of protection regimes can help us explain what is happening in their other 'formal' side. More specifically, the paper provides us with the resources to move beyond conventional approaches that suggest that formal rights and liberal advocacy is all that protects migrants from violence. The concept of 'migrant protection regimes' compels us to observe 106 NGO nos 5, 19; Official nos 5, 15. This analysis of the second law does not comport with the interpretations offered by all sources. Some NGO workers believed the junta was manipulating the process the entire way, and simply using the threat of big finesand mass exodusto force employers to register their workers (for example, NGO no. 18). However, the accounts of officials inside the Labour Ministry do not support this line of interpretation, but ratherin the words of one official, support the image of a government 'learning by doing' (Official no. 12) and being countered and hemmed in by the responses of providers and seekers of protection along the way. the full gamut of protection relationships that employers, policymakers, officials, and ordinary migrants seek to establish, renegotiate, and break, and to reckon with the situated, interactive, and non-linear way in which agents change these regimes.
In contrast to conventional IR scholarship, this approach demands that we look beyond the weak, formal dimensions of the international migration regime to understand how migrants are governed. This study concurs with previous research, that these informal dimensions of the international migration regime tend to result in migrant exclusion and exploitation, and that 'protection' is a racket heavily skewed against those who do not possess violent means. At the same time, this study also exposes the significant collective leverage that migrants can wield to resist or respond to the monopolistic power grabs and exploitative strategies of protection providers, by 'exiting' a protection regime.
While this study is based on a single case study of a specific moment in one country, there are several reasons to believe why the migrant protection regime concept might offer fruitful lines of inquiry beyond Thailand in 2017. On the one hand, this research has the potential to add significant purchase to efforts to understand migrant protection for the largest share of the world's migrants, who move between countries located in the Global South, where formal rights and liberal advocacy tend to be weaker and violent and informal modes of governing comparatively strong. However, the concept also has potential to travel further than that. As refugee protection systems in the Global North continue to fragment, governments experiment with guest workertype policies that offer migrants fewer rights, and informal migration remains a structural feature of most national economies, migrants in the Global North will continue to navigate between formal and informal protection relationships. Understanding how protection providers and seekers attempt to shape and respond to changes in informal migrant protection regimes, may therefore offer key insights into the nature and transformation of emerging migration regimes across the Global North and South.
Going beyond the study of international migration, this work opens pathways to rethink the agency of vulnerable populations in IR. Historical sociologist and feminist research has popularised a deeply critical understanding of sovereign protection as a 'racket' that tips the scales in the favour of violent states and men and against civilians and women. This article demonstrates the value of using this critical take on the social construction of protection, to explore how migrant protection is achieved in practice. At the same time, it reveals the unique potential that vulnerable populations have to change the terms of protection on offer. Here, Hirschman's concept of 'exit' helps us point to the fact thatwhile protection may be a universal racket, human mobility can be a dynamic force, creating demand for change in the terms upon which protection is offered. 109 The challenge for researchers is to discover how the political meaning of such powerful acts of mobility might be interpreted and translated into more open protection regimes.