The Turkish Armed Forces, together with the Syrian National Army, just launched #OperationPeaceSpring against PKK/YPG and Daesh terrorists in northern Syria […] #OperationPeaceSpring will neutralize terror threats against Turkey and lead to the establishment of a safe zone, facilitating the return of Syrian refugees to their homes.Footnote 1
Although there is no accepted legal definition of a ‘safe zone’,Footnote 2 the term generally refers to the establishment of areas where civilians may find refuge from armed conflict.Footnote 3 Safe zones can take many forms, but often consist of camps in refugee-generating States that are protected by the military power of a foreign State or by an international organisation. Safe zones are worth taking seriously for several reasons. The image of three-year-old Alan Kurdi lying face down on a Turkish beach represents just one of the thousands of people who have died trying to leave Syria in recent years to claim asylum in Europe.Footnote 4 The creation of a safe zone could allow individuals to avoid these perilous journeys by seeking protection in their own State. A safe zone might also be the only feasible protection option where borders are closed, where flight is impossible, for those who do not wish to leave their State, or—as stated by President Erdoğan—to allow those who have left to return. As the numbers of displaced persons generated by the Syrian conflict rises and the political will to accept refugees falls, it is unsurprising that Turkey has established a safe zone in Syria and that this course of action was also previously considered by the US, Iraq, France and Russia.Footnote 5
However, safe zones can be extremely dangerous. As is clear from President Erdoğan's comment, circumvention of refugee flows is often their main objective.Footnote 6 The belief that adequate protection may be found in a safe zone could tempt governments of asylum States to create them, possibly placing more emphasis on containment than on the rights of individuals to leave and seek asylum. In addition, safe zones are often unworkable.Footnote 7 It is difficult, if not impossible to secure the consent of all parties to the conflict for the establishment of the zone, and consequently it is difficult to ensure the zone's protection from attack. This was tragically illustrated by the 1995 example of Srebrenica. Despite its status as a UN-designated safe zone, Srebrenica fell to the Bosnian Serb forces, resulting in the deaths of over 7,000 Bosnian Muslims.Footnote 8 Further criticisms of the safe zone established by Turkey in Syria include that it has been established illegally,Footnote 9 that it could potentially destabilise the region,Footnote 10 and that it is directed at ethnically cleansing the Kurdish population in northern Syria.Footnote 11
This article focuses on the first ‘danger’ of creating safe zones, namely whether in international refugee law, safe zones can be invoked as a ground for refusing asylum. It examines the concept of the Internal Protection Alternative (IPA),Footnote 12 which posits that an individual is not a refugee if there is a safe place within his/her country where he/she can relocate.Footnote 13 Examples of IPAs could include relocating from the countryside to the city where an individual is less likely to be found by his or her persecutors;Footnote 14 or relocating to an area where a clan, tribe, militia, or international organisation could provide protection.Footnote 15 The IPA concept is central to this article because many States claim that safe zones can qualify as alternatives for asylum and thus as IPAs.Footnote 16 This article will take as its starting point a situation where an individual's refugee status is being assessed and will determine whether the existence of a safe zone in the individual's country of nationality could qualify as an IPA.
This article recognises that although safe zones can be deeply problematic, scholars have a crucial role to play in responding to the intentions of States. Thus it is important to set out how safe zones interact with international refugee law. This article will distil the binding minimum criteria when determining whether an IPA exists which consequently must be applied by those States who wish to use safe zones as an alternative to recognising refugee status.Footnote 17 These binding minimum criteria are that (i) the proposed IPA must be accessible to the applicant; (ii) there is no risk of exposure to the original risk of persecution in the proposed IPA; and (iii) there must be no new risk of persecution or of refoulement in the proposed IPA.Footnote 18 These criteria are cumulative. Therefore, a failure to meet any one of these criteria means that there is not a valid IPA. It is worth noting that this author does not necessarily agree that this low threshold should be the yardstick by which a lawful IPA is measured. However, this author accepts that the terms of the treaty itself and the actual practice of States, as set out in the 1969 Vienna Convention on the Law of Treaties,Footnote 19 are amongst the key components for establishing the minimum binding criteria applicable in the IPA inquiry.
The article then engages in an innovative historical analysis to assess whether safe zones can satisfy the IPA criteria. It will examine three historical case studies—Sri Lanka, Srebrenica, and Iraq—which were selected primarily on the basis that their respective conditions were frequently studied by historians and political scientists.Footnote 20 Hence nowadays there is significant information available about the conditions within the zones. By applying the IPA criteria retrospectively to these case studies, this article argues that safe zones can qualify as IPAs but only in exceptional circumstances.
The structure of the article will be as follows. This section (section I) sets out the article's aims and section II will set out its contribution to literature. Section III will introduce the safe zone concept against the broader background of refugee containment. It will assess the benefits and pitfalls of safe zones and will trace the history of the establishment of safe zones. This will lay the foundation for the introduction of the IPA concept in section IV. Section IV will explain that in order to be an alternative for refugee status, in law, a safe zone must satisfy the criteria for a valid IPA. Using the rules of treaty interpretation, section IV will illustrate what these criteria are. Section V will examine three case studies of safe zones to test whether they satisfied the IPA criteria, and to draw general lessons on whether a safe zone can constitute an IPA. Section VI will conclude by summarising the article's key claims and setting out its implications for the potential future establishment of safe zones.
II. SAFE ZONES: MINDING THE GAP IN SCHOLARSHIP
To date, the literature on safe zones has mainly focused on the politics surrounding their establishment.Footnote 21 Very little literature examines the legal framework pertaining to safe zones,Footnote 22 and the relationship between international refugee law and safe zones has not been studied at all. This is despite the widely held view that safe zones may undermine the institution of asylum and despite calls for further research on the relationship between refugee law and safe zones.Footnote 23 This article fills this gap. In addition, this article contributes to the literature on the IPA more generally. From 1998 to 2003, most of the literature on the IPA focused on how to accommodate the IPA within the confines of the 1951 Refugee Convention.Footnote 24 However most, if not all of these pieces do not reflect the law as it stands and rather put forward a normative position regarding which criteria should be applicable. For example, some authors consider that durability and/or reasonableness and/or protection of human rights standards are stand-alone criteria in the IPA inquiry but these criteria do not represent widespread State practice.Footnote 25 Moreover, it is unclear how these additional criteria should be interpreted, and whether they have any added value when compared to those criteria which are already clearly established in State practice. As a result of the lack of clarity in the literature to date about what the binding applicable standards are in the IPA analysis, there is a risk that asylum-seekers will be returned returned to their countries of origin without adequate safeguards.Footnote 26 By surveying widespread State practice and the rules of treaty application, this article identifies the three binding criteria that apply to the IPA analysis under international law and will thus be of relevance to both the safe zone context and also when applying the IPA more generally.
Consequently, this piece brings together two strands of parallel literature—the political science literature on safe zones and the refugee law literature on the IPA—to examine, for the first time, the applicability of the IPA criteria to safe zones. This analysis is long overdue considering that safe zones have been employed as a foreign policy tool since the nineteeenth century,Footnote 27 and have been recently established in Sri Lanka (1990), Iraq (1991), Rwanda (1994), Bosnia (1993–1995), Afghanistan (2000–2001), Somalia (2007) and Syria (2019). The fact that the application of the IPA will probably become mandatory for EU States makes the invocation of safe zones as an IPA all the more likely.Footnote 28 Thus this article puts the practice of States under the microscope, seeks to establish whether past safe zones were lawful alternatives to asylum and in doing so, may inform States’ future plans in respect of establishing safe zones as alternative to asylum.Footnote 29
Having set out this article's main contribution to scholarship, the next section (section III) will illustrate how State hostility towards refugees has led to the implementation of increasingly creative policies of deterrence, including the establishment of safe zones in refugee-generating States. The analysis that follows in the remainder of the article will be set against this political background.
III. A PARADIGM SHIFT: FROM COMPASSION TO CONTAINMENT
Refugee policy has always been partially based on compassion and partially based on State interest. The trend of establishing safe zones therefore needs to be assessed against the broader political context of refugee containment, deterrence, and deflection.
Before the end of the Cold War, the recognition of refugees was viewed as a strategic political act highlighting the failure of a State to protect its nationals.Footnote 30 Accepting refugees could also serve the strategic aims of host countries by destabilising the refugees’ home countries.Footnote 31 However, in the post-Cold war era, the interests of refugees and refugee-receiving States no longer overlapped and refugees began to lose their political currency.Footnote 32 As refugee numbers grew, and came from further afield, States began to employ creative means to constrain refugee flows and restrict the number of individuals they recognised as refugees. They did so in four main ways. First, they curtailed the entry of refugees onto their territories through relatively invisible —and hence politically expedient —non-entrée measures.Footnote 33 Carrier sanctions were introduced, walls and fences were built, visas were denied, first country of asylum principles were applied, refugee applications were processed offshore, and boatloads of refugees were pushed back at sea. At the time of writing, the US is constructing a ‘new border wall system’ along its border with Mexico;Footnote 34 a case has been filed in the European Court of Human Rights challenging Italy's pushback policy of migrants at sea;Footnote 35 and Australia has controversially repealed a law which allowed sick refugees held offshore to be treated in Australia.Footnote 36 With the rise of right-wing politics in many asylum States, governments are likely to become even more hostile to asylum seekers and find increasingly creative ways to restrict refugee flows.
Second, those who manage to reach the territory of asylum States tend to receive very little by way of assistance. Reception conditions for asylum-seekers frequently fall below international human rights law standards, which forms a further hurdle in accessing asylum. As the recent practice of the US illustrates,Footnote 37 asylum-seekers are often placed in detention or camps with limited freedom of movement while they wait for their claim to be processed.Footnote 38 They often have no right to work and are segregated from the community.Footnote 39 The obligation of the State to hear an asylum-seeker's claim is sometimes outsourced to private actors and asylum-seekers frequently report difficulties in accessing legal aid and fair and transparent asylum determination proceedings.Footnote 40 All of these hurdles make it difficult in practice for the asylum-seeker to put their best case forward and consequently makes it easier for the State to reject their claim.
Third, States increasingly interpret the 1951 Refugee Convention to suit their interests of limiting their refugee intake. States routinely reject asylum claims on the basis that there is an alternative safe place within the country of origin to which the individual could relocate. As will be explained in more detail in section IV, this principle became known as the IPA and is now well entrenched in the jurisprudence of many refugee-receiving States. States also regularly reject asylum applications on the grounds that protection is available from non-State actors of protection such as clans, tribes, international organisations, and militias.Footnote 41 Certain States are presumed to be ‘safe’ which means that asylum claims from nationals of these States were highly likely to be rejected.Footnote 42 In addition, States have concluded agreements to send refugees and asylum-seekers to so-called ‘safe’ countries—an example of which would be the recently concluded Turkey–EU agreement.Footnote 43
Fourth, the attention of the international community has turned to in-country protection, as demonstrated by the increased focus on the plight of Internally Displaced Persons (IDPs). In recent years, the completion of the Guiding Principles on Internal Displacement and regional IDP treaties,Footnote 44 the appointment of a UN Special Rapporteur on the Human Rights of IDPs, the involvement of the United Nations High Commissioner for Refugees (‘UNHCR’) with IDPs, the development of the principles of ‘Responsibility to Protect’Footnote 45 and ‘Humanitarian Intervention’,Footnote 46 and the recent establishment by the UN Secretary-General of a High-Level Panel on Internal Displacement all illustrate a growing willingness of the international community to protect those displaced within borders.Footnote 47 However, these developments were not exclusively humanitarian in nature. For example, some authors have suggested that the NATO bombing during the 1990s Kosovo crisis was motivated at least in part by States’ unwillingness to accept large numbers of refugees.Footnote 48 These motivations also contributed to promotion of terms such as ‘safe zones’, ‘buffer zones’, the ‘right to remain’, and ‘preventative protection’. In addition, foreign-imposed safe zones were established as an alternative to asylum.Footnote 49
The concept of safe zones is controversial for a long list of reasons, an appraisal of which is outside the scope of this article. The most important reason for the purposes of this piece is that safe zones could threaten the institution of asylum. There is a danger that refugee-receiving States could (mis)interpret safe zones as an IPA and therefore as an alternative for asylum. This is particularly relevant for UNHCR's involvement with displaced persons in their countries of origin. UNHCR is likely to play a key role in the establishment of any future zones as it has particular responsibility at a global level for protection, shelter and camp coordination and management for conflict-related IDPs.Footnote 50 Understanding how the protection of individuals in a safe zone interacts with the IPA concept is of particular importance for UNHCR because if safe zones are interpreted as an alternative for asylum, UNHCR could be undermining its primary mandate of refugee protection.Footnote 51
The establishment of safe zones in countries of origin therefore forms part of a general trend of containment policies as identified above. As well as invoking a safe zone as an IPA (and thus refusing asylum applications) or ceasing the refugee status of individuals they have previously recognised as refugees,Footnote 52 States could argue in future that because of the existence of a safe zone in the country of origin, they are justified in closing borders, imposing carrier sanctions, pushing back boatloads of refugees, refusing to issue visas to nationals of that country. They could argue that such actions are lawful because the availability of safe zones means the individuals are not refugees, and thus are not protected by the prohibition of refoulement. The issue of whether a safe zone is an IPA may therefore not even come before a refugee status determination body, as individuals may not even make it out of their respective States in order to claim asylum. This makes the conditions in the safe zones less visible to the international community and States who establish safe zones for the purposes of limiting refugee numbers consequently perceive and/or portray themselves as less accountable for their actions. This in turn gives them even less incentive to ensure that safe zones satisfy the minimum IPA criteria. Thus, in this article, the term ‘asylum’ is used rather than ‘international protection’ to highlight the fact that IPAs are not only used as an alternative for international protection, but also to deny individuals the right to leave their country and access asylum proceedings.
Underlying these creative attempts to restrict obligations towards refugees is an asylum system that, with each passing year, is under increasing pressure. Developing regions hosted 84 per cent of the world's refugees in 2018 and the main States of asylum—such as Turkey, Pakistan, Uganda, and Lebanon—frequently voice that they are struggling to cope with the refugee influx.Footnote 53 The recently concluded Global Compact on Refugees (GCR) aims to encourage responsibility sharing but as a non-binding agreement, the GCR is not legally opposable to States and relies on their goodwill for its successful implementation.Footnote 54 In the likely event that developing regions continue to struggle to manage their refugee populations, we may see their generosity waning and a corresponding increase in border closures and refoulement. Moreover, although Western States host significantly less refugees than their counterparts in the developing world, the Western world has seen a rise in right-wing politics and corresponding references to refugee ‘crises’ on their territories.Footnote 55 The use of the term ‘crisis’ and its corresponding restrictive migration policies draws attention away from the problem in the countries of origin, strips decision-makers of responsibility, and presents certain decisions as necessary.Footnote 56 As the asylum system moves towards breaking point, it is unsurprising that Turkey has established a safe zone in Syria.
It is therefore of crucial importance that we determine whether the latest effort to contain refugee flows via the establishment of safe zones is in fact compatible with refugee law. The following section (IV) will set out the relationship between the concept of safe zones and refugee law—notably, that a safe zone may be invoked as a potential IPA.
IV. WHAT IS AN INTERNAL PROTECTION ALTERNATIVE?
A refugee is an individual who, owing to a well-founded fear of persecution on specific grounds, is outside their country of nationality and is unable or unwilling to return to it because of lack of available protection.Footnote 57 The IPA concept means that an individual is not a refugee if there is a safe place within his/her country of nationality where he/she can relocate.Footnote 58 The concept is known by many names, including ‘Internal Protection Alternative’, ‘Internal Protection Principle’, ‘Safe Haven Principle’, ‘Internal Resettlement’ and ‘Internal Relocation Alternative’. This article will use the term ‘Internal Protection Alternative/IPA’ because it appropriately focuses on the protection (or lack thereof) in the country of nationality, which is a criterion of the refugee definition. The IPA concept is central to this article because some States are claiming that safe zones can qualify as IPAs, and thus as alternatives for asylum.Footnote 59 As aforementioned, this claim has not yet been subject to scrutiny in the literature.
Although the IPA concept does not explicitly appear in the 1951 Refugee Convention, the existence of a putative IPA is often considered as part of the inquiry as to (i) the existence of a well-founded fear of persecution and/or (ii) lack of protection from persecution as set out in the 1951 Refugee Convention. Put simply, a person cannot be said to be ‘owing to a well-founded fear’ unable or ‘unwilling to avail himself of the protection of that country’ if he or she has access to the protection of that State, albeit in a different part.Footnote 60 Thus, the IPA concept is generally accepted to be compatible with the 1951 Refugee Convention,Footnote 61 and is applied widely by States.Footnote 62
The fact that the IPA is not explicitly anchored in the 1951 Refugee Convention has led to wide divergence in the IPA's interpretation by States. This divergence is compounded by the fact that the 1951 Refugee Convention does not have a central binding monitoring or enforcement mechanism, which can deliver authoritative interpretations of the 1951 Refugee Convention. The criteria applied by States falls into two categories: (i) a handful of criteria that are applied in a widespread manner and represent the legally binding minimum standard;Footnote 63 and (ii) a long list of criteria which are applied by some States but do not have widespread acceptance, and thus do not yet have the status of binding international law.Footnote 64 As explained in the Introduction, this article will take a pragmatic approach by using the rules on treaty interpretation to derive the binding minimum criteria in the IPA inquiry and will then apply these criteria to safe zones.
A note on methodology is helpful at this stage to explain how these criteria are derived. As aforementioned, the IPA is often considered to be an implicit part of the 1951 Refugee Convention. The 1969 Vienna Convention on the Law of Treaties (VCLT) provides that a treaty shall be interpreted in good faith in accordance with the ordinary meaning to be given to the terms of the treaty in their context and in the light of the treaty's object and purpose.Footnote 65 The first point of departure for determining the criteria for the IPA (and hence how the refugee definition is interpreted) is therefore the terms of the refugee definition itself as found in the 1951 Refugee Convention. In addition, ‘subsequent practice’ should be taken into account when determining the meaning of a treaty.Footnote 66 Both judgments of domestic courts and the practice of regional systems created by States (such as the EU) constitute subsequent practice for the purposes of interpreting the IPA.Footnote 67 Finally, the VCLT also directs us to take into account ‘other relevant rules of international law’ when interpreting the meaning of a treaty.Footnote 68 Thus understanding how the IPA has been interpreted in other treaty regimes also helps us to understand how it should be interpreted in relation to the 1951 Refugee Convention.
The following sections will set out the three criteria that are generally accepted as forming part of the IPA inquiry: (i) the IPA must be accessible to the applicant; (ii) there must be effective protection from the original risk of persecution; and (iii) there must be no new risk of being persecuted or of refoulement. As the following sections will show, these three criteria derive from the ordinary meaning of the terms of the 1951 Refugee Convention and are widely applied in State practice.
A. The IPA Must Be Accessible to the Applicant
First, the IPA must be safely and legally accessible to the applicant.Footnote 69 In the words of the Canadian Federal Court of Appeal:
An [internal protection alternative] cannot be speculative or theoretical only; it must be a realistic attainable option. Essentially, this means that the alternative place of safety must be realistically accessible to the claimant. Any barriers to getting there should be reasonably surmountable. The claimant cannot be required to encounter great physical danger or to undergo undue hardship in travelling there or in staying there. For example, claimants should not be required to cross battle lines where fighting is going on at great risk to their lives in order to reach a place of safety.Footnote 70
The requirement that the IPA must be accessible flows from a ‘good faith’ interpretation of the 1951 Refugee Convention.Footnote 71 If the proposed IPA is not accessible, the applicant cannot be expected to seek protection from persecution there and the applicant may qualify for international protection (ie they may be recognised as a refugee).
B. There Must Be Effective Protection from the Original Risk of Persecution
The second criterion—that there is effective protection from the original risk of persecution—essentially means that the relocation area must be an ‘antidote’Footnote 72 to the persecution feared.Footnote 73 This criterion is uncontroversial because it is anchored in the text of the 1951 Refugee Convention. Article 1A provides that an individual is a refugee, inter alia, if he/she has a well-founded fear of persecution. It is generally accepted that if the fear of persecution can be remedied by relocating within the State, the individual does not qualify as a refugee.Footnote 74 However, if relocation would expose the individual to persecution, the area of relocation cannot be said to provide an antidote to the persecution feared and thus cannot qualify as an IPA.
There is no definition of what constitutes persecution in the 1951 Refugee Convention. Indeed, it is generally accepted that the drafters of the 1951 Refugee Convention intentionally declined to define ‘being persecuted’ because they recognised the impossibility of exhaustively listing all the forms of maltreatment that might entitle persons to refugee status.Footnote 75 However, many different scholarly authorities propose a human rights interpretation of the term ‘persecution’,Footnote 76 and this approach is generally accepted in State practice.Footnote 77 It is nonetheless important to note that not all human rights violations will necessarily constitute persecution. Assessment will turn on factors including (i) the nature of the freedom threatened; (ii) the nature and severity of the restriction;Footnote 78 and (iii) the likelihood of the restriction eventuating in the individual case.Footnote 79 Examples of serious harm sufficient to ground a risk of being persecuted may include a risk to life, the risk of torture or cruel, inhuman or degrading treatment or punishment, slavery, risk of beatings or other physical violence, and the risk of violation of socio-economic rights.Footnote 80
C. There Must Be No New Risk of Being Persecuted or of Refoulement
The third generally accepted criterion is that there must be no new risk of being persecuted or of refoulement.Footnote 81 This criterion is uncontroversial because, as set out above, lack of protection from persecution is one of the criteria for establishing refugee status, and refoulement is prohibited by Article 33(1) of the 1951 Refugee Convention. Although establishing whether there is a risk of persecution and/or refoulement is a highly complex inquiry,Footnote 82 in essence this criterion prohibits the expulsion or return of any individual who could face a new risk of persecution, or where there are substantial grounds for believing that he or she would be in danger of torture, inhuman or degrading treatment or punishment if returned to a particular IPA.Footnote 83 In addition, there must be no risk of ‘indirect’ refoulement ie, an applicant should not be returned to an inadequate IPA where the conditions there are so unreasonable that an applicant may in desperation return to the territory of her persecutors or another area where there is a real risk of serious harm.Footnote 84 For example, poor socio-economic conditions such as lack of food, shelter, housing, and/ or medical care—even if they do not amount to persecution—could rule out an area from being a lawful IPA.
D. Examining the Criteria
It can be very challenging to establish whether these three criteria are satisfied. The inquiry as to whether an individual is a refugee is forward-looking in nature but it is not clear how far forward the assessment may extend and to what extent and whether the harm feared must be imminent.Footnote 85 Moreover, it is very difficult to predict whether a proposed IPA is in fact safe and/or accessible, and whether it will remain so. This difficulty is compounded when the proposed IPA is a safe zone. A safe zone is generally located in an area engulfed in armed conflict. The nature of warfare means that the security situation will change day to day, perhaps hour by hour; and receiving accurate information can be almost impossible. Whereas asylum determination bodies can usually rely on reports from media, NGOs, academic experts and international organisations to determine conditions in the proposed IPA, such information simply is not available or is out of date or inaccurate during armed conflict. It may be too dangerous for individuals to report from the conflict zone, and/or the communication of accurate information may be hampered by parties to the conflict or by the effects of warfare. Moreover, a cynical view might posit that States who create and monitor safe zones as an alternative for refugee status simply do not have any incentive to publicise safety shortfalls in the zone, as such information could then be relied upon in refugee status determination proceedings and lead to higher numbers of recognised refugees. Similarly, international organisations that rely on States for funding could be tempted to under-report. For example, the UNHCR whose very mandate obliges it to be impartial, has been accused of misrepresenting the situation in a safe zone in Sri Lanka.Footnote 86 Indeed, such under-reporting is to be expected, given that safe zones are often created out of motivation to stop refugee flows rather than a motivation to provide superior protection.Footnote 87 Thus even though the baseline criteria for the IPA analysis can be clearly identified, it is difficult—particularly in the context of safe zones—to determine whether these criteria are actually satisfied.
Having set out the criteria applicable to the IPA analysis, the following section will examine how these criteria might apply to specific scenarios.
V. CAN A SAFE ZONE EVER BE AN INTERNAL PROTECTION ALTERNATIVE? CASE STUDIES OF SRI LANKA, IRAQ AND SREBRENICA
As aforementioned, the practice of creating safe zones has been firmly established in State practice since the nineteenth century and has come to recent prominence with the establishment of safe zones in Sri Lanka (1990), Iraq (1991), Rwanda (1994), Bosnia (1993–1995), Afghanistan (2000–2001), and Somalia (2007) and Syria (2019).Footnote 88 An analysis of all these safe zones is beyond the parameters of this article and thus the examination will be limited to the cases of Sri Lanka (1990), Iraq (1991) and Srebrenica (1995). These three scenarios are particularly suitable to determine the relationship between the IPA criteria and safe zones because: (1) It would normally be necessary to engage in extensive fieldwork to assess the conditions within the respective safe zones. However, in the years following the establishment of these three selected safe zones, their respective conditions were frequently studied by historians and political scientists. Hence nowadays there is significant information available about the conditions within the zones and it is thus possible to assess whether they fulfilled the IPA criteria. (2) This time period (the 1990s) is helpful to analyse new developments because it marked the beginning of the general trend of containment, deterrence, and deflection of refugees which continues today.Footnote 89 (3) The case studies chosen represent a diversity of actors of protection and hence possible forms of protection—States (Iraq); international organisations (Sri Lanka); and States acting under the mandate of an international organisation (Srebrenica). (4) The safe zones chosen were established in different ways: legally, with the consent of the host State (Sri Lanka); illegally, without the consent of the host State (at least initially) (Iraq); and legally, with the consent of the host State and pursuant to Security Council authorisation (Srebrenica).Footnote 90 Hence the case studies chosen shed some light on whether legality affects satisfaction of the IPA criteria. (5) The scenarios chosen represent different geographic areas—Europe, South Asia, and the Middle East. (6) The scenarios chosen have been perceived by scholars as having varying degrees of success in terms of providing protection from persecution. Sri Lanka and Operation Provide Comfort have been perceived as being relatively successful, and Srebrenica has largely been perceived to be a failure. Consequently, the samples chosen represent a diversity of circumstances and outcomes, which in turn will lay the foundation for a rigorous analysis of whether a safe zone can constitute an IPA.
Three notes of clarification are useful before turning to the individual scenarios. First, it is important to note that even if a particular safe zone satisfies all three of the culmulative ‘baseline’ criteria, that is not to say that all States would view that particular safe zone as a viable IPA. As aforementioned, a large number of refugee-receiving States apply additional criteria to the IPA inquiry and thus although a safe zone might satisfy the three criteria identified in this article as binding under international law, they could fail other criteria as applied by individual States which are not internationally binding. Second, in the 1990s—the time period from which the scenarios in this article were chosen—there was no clearly established consensus on the applicable IPA criteria. Thus the criteria are being examined retroactively to the selected scenarios. Refugee status determination is always forward-looking but identifying whether past safe zones might have satisfied the IPA test may help us understand whether the establishment of any future safe zones would satisfy the IPA test. Third, the analysis of whether each of the respective safe zones qualifies as an IPA is done at a general level. The question in essence is whether the conditions in the zone are such that the return of any individual there would be a violation of the 1951 Refugee Convention. It does not take into account individual circumstances (such as the reason for flight, gender, health etc) which may be highly pertinent in refugee status determination proceedings and lead to a more restrictive understanding of what qualifies as a lawful IPA.Footnote 91
A. Sri Lanka (1990)
The Eelam War is the name given to the second phase of armed conflict between the Sri Lankan Military and the separatist Liberation Tigers of Tamil Eelam (LTTE). This war began in 1990, after India had withdrawn from Sri Lanka and peace talks between the government and the LTTE.Footnote 92 At that time, the UNHCR had been close to completing a returnee operation in Sri Lanka. India was unwilling to host the refugees, and the UNHCR was thus faced with the choice of abandoning those receiving reintegration assistance or to look after them and in doing so, reduce the pressure to leave.Footnote 93 In November 1990, Open Relief Centres (ORCs) in Sri Lanka were established, which were temporary places where displaced persons on the move could freely enter or leave and obtain essential relief assistance in a relatively safe environment.Footnote 94 Their establishment can be distinguished from other safe zones for three reasons. First, the ORCs were negotiated on the ground, rather than forcibly established by other States operating through the Security Council and there was consensus between the parties to the conflict that it was a safe area.Footnote 95 Second, ORCs were situated in both LTTE and government-controlled areas, and were not associated with territorial claims.Footnote 96 Third, the ORCs were demilitarised.Footnote 97
The analysis now turns to whether the ORCs satisfied the IPA criteria. All of the literature examined for the purposes of this article did not mention any hurdles to accessing the IPA.Footnote 98 Thus it may be presumed that the first criterion—that of accessibility—was satisfied. This is supported by the fact that many individuals sought protection within the ORC.
It is unclear whether the other two criteria for a valid IPA are satisfied, notably the requirements of effective protection from the original risk of persecution, and no risk of new persecution or refoulement. In terms of lives saved most commentators have deemed the ORCs to be a success.Footnote 99 The majority of studies on the ORCs report that they operated relatively safely, despite occasional evictions and security incidents.Footnote 100 Tiso praises the ORCs highly, noting that by February 1991, more than 50,000 Sri Lankans were residing in the ORCs and the ORCs also enabled people who were living in nearby areas to temporarily seek refuge in the ORCs when threatened.Footnote 101 He also remarked that ‘the United Nations should build on the Sri Lankan experience as the operation has established many valuable precedents that the United Nations should follow in future crisis situations’.Footnote 102 The UNHCR described the ORCs as ‘havens of safety, accepted and respected by both warring parties’Footnote 103 and that they guaranteed ‘not only better living conditions, but also protection’.Footnote 104 Many Sri Lankans returned home and several States in Europe and Asia signed agreements that provided for voluntary repatriation of Sri Lankan refugees.Footnote 105 In addition, Tamils were not debarred from seeking asylum outside the country because of the existence of the safe areas.Footnote 106
However, Arulnatham and the Immigration and Refugee Board of Canada have reported that many Tamil refugees fled the ORCs because the Sri Lankan government had seized individuals from within the camps, many of whom were tortured or never returned.Footnote 107 Reportedly, although the UNHCR's officials attempted to recover these people, the camp's director told the refugees that UNHCR could no longer guarantee their safety while in the camp. Nonetheless, UNHCR published a report stating that the camp was open and providing protection and relief for those seeking shelter from the conflict, making no reference to the protection failures in the camp. This is particularly concerning because if the ORC was ineffective, publicising this information was crucial to ensure that individuals could receive asylum elsewhere. Moreover, many individuals risked their lives to travel to the ORC as they believed that it was safe. The fact that the UNHCR is funded by governments and needs government support in order for it to operate within countries of origin may have affected its willingness to report accurately about the situation on the ground. In order to assess whether safe zones are an alternative for refugee status, it is crucial that the conditions in the zone are accurately reported.
In sum, it is unlikely that the ORCs qualified as IPAs. Even though they were accessible, there were mixed reports concerning their safety. The three IPA criteria are cumulative. Therefore, a failure to meet any one of these criteria means that there is not a valid IPA.
B. Iraq (1991) (Operation Provide Comfort)
Saddam Hussein's belligerent actions in the late 1980s and early 1990s led to significant fear among the Kurdish population in Iraq, particularly after he launched a chemical attack on the Kurdish town of Halabjah in 1988.Footnote 108 Kurdish rebels attempted to overthrow the government in 1991, having been encouraged to do so by US President George Bush.Footnote 109 The rebellion failed and thousands fled fearing reprisals from Iraq.Footnote 110 Soon after, Turkey closed its borders because it viewed the Kurdish refugees as a political problem.Footnote 111 It relied on its geographic limitation to the 1951 Refugee Convention, maintaining that it was only obliged to accept as refugees those having a well-founded fear of persecution as a result of events occurring in Europe.Footnote 112 At that time, there were more than 200,000 Kurdish individuals trapped in the mountains of the Turkish–Iraqi border, living in dire conditions.Footnote 113
The initial response of the UNHCR was to attempt to persuade Turkey to open its borders.Footnote 114 However, the strategic importance of the Turkish border to the United States, United Kingdom, and France meant that within days, UNHCR's efforts were hampered by the inter-State negotiations that reflected United States’ and allies’ reluctance to condemn the Turkish border closure.Footnote 115 Notwithstanding the fact that Security Council Resolution 688 reaffirmed the ‘sovereignty, territorial integrity, and political independence of Iraq and of all States in the area’, the United States, United Kingdom, and France interpreted the resolution's reference to the ‘massive flow of refugees towards and across international frontiers’ as justifying the creation of a safe zone for Kurds in Iraq without Iraqi consent, which they termed ‘Operation Provide Comfort’ (OPC).Footnote 116 The safe zone's compatibility with Iraq's territorial integrity was emphasised in order to allay the fear that humanitarian action might encourage Kurdish self-determination and thereby destabilise Iraq and the region. The safe zone was thus not for permanent settlement but a set of ‘revolving doors, with refugees leaving to return home while others take their place’.Footnote 117 It would only be large enough to provide a temporary shelter until the displaced felt secure enough to go home. The operation was aimed at being short-term because the leaders wished from the very beginning to hand over the operation to the UN as soon as possible.Footnote 118 On 16 April 1991, the allied powers announced their decision to send troops to the border area to set up and administer camps. Six camps were set up, each to shelter between 60,000 to 65,000 people.Footnote 119 A no-fly zone was also set up to prevent further attacks on the displaced persons.Footnote 120
For Iraq, creating a safe zone without consent was a clear breach of its sovereignty. To avoid the humiliation this would entail, Prince Sadruddin Aga Khan—the personal representative of the UN Secretary-General for Humanitarian Assistance Related to the Iraq-Kuwait Crisis—signed a memorandum of understanding on 18 April 1991 with the Iraqi government allowing for the establishment of UN humanitarian centres within the country.Footnote 121 The memorandum emphasised that the operation was not directed at inspiring Kurdish self-determination and respected Iraq's territorial sovereignty. Security would be provided only for the personal safety of IDPs and as a temporary measure. Towards the end of April 1991, efforts were directed to the prospect for a mass ‘voluntary’ repatriation of Kurds from border areas, and UNHCR agreed to participate in the operation as a ‘reluctant but pragmatic partner’.Footnote 122 The allied haven doubled its size and the UN subsequently took over the administration of the camps.Footnote 123
There is a general consensus that the safe zone established by OPC was a success.Footnote 124 In terms of the IPA criteria, it seems as though it satisfied criteria of accessibility as it was accessible to those for whom it was meant to protect—IDPs in the mountainous border with Turkey. Insofar as this author's research has revealed, there are no reports of a risk of persecution and/or refoulement within the zone. Goodwin-Gill and McAdam cautiously leave open the possibility that returning an asylum seeker to these safe zones may have been compatible with the 1951 Refugee Convention.Footnote 125 It thus seems as though the three baseline criteria were satisfied in relation to OPC and that OPC satisfied the lowest common denominator of IPA criteria applied by States.
At this point it is apt to assess whether OPC might be viewed as an IPA even by those States with more generous interpretations of the 1951 Refugee Convention and who apply the IPA more sparingly. A detailed analysis of all criteria applied by such States is outside the scope of this article but suffice it to say that at first glance,Footnote 126 it appears that OPC would satisfy many such criteria. For example, although some States apply a presumption that there is no IPA where the agent of persecution is the State, this presumption can be rebutted where there is compelling evidence of the ability to deliver durable protection in the proposed IPA.Footnote 127 In this scenario, the threat of reprisals from the allied forces was credible, and Iraq seemed to be cooperating with their demands. Therefore, the fact that Iraq was the agent of persecution would not necessarily preclude OPC from being a lawful IPA. Moreover, some States interpret the term ‘effective protection’Footnote 128 as meaning that the risk of persecution is ‘eradicated’.Footnote 129 Hindsight informs us that the basis of the IDPs’ and refugees’ fear—reprisals from Saddam Hussein's government—was in fact eradicated by OPC.
In sum, OPC was likely to have satisfied the baseline criteria for a lawful IPA and may even have satisfied the higher threshold applied by some States.Footnote 130 However, invoking OPC as a precedent for the creation of future safe zones should be done with caution. This is because the conditions for a successful safe zone are rarely as good as they were in northern Iraq. Iraqi military forces were badly damaged following the Gulf War. Iraqi ground forces had to cross much open country to attack the Kurds and they were very vulnerable to air attack.Footnote 131 The Kurds had strong fighting abilities, and it was clear that the US had an incentive to protect the Kurds so as not to take away from the Desert Storm victory.Footnote 132 In short, the conditions were unusually favourable to the establishment of a safe zone.
C. Srebrenica (1993–1995)
As the former Yugoslavia began to disintegrate, the Republic of Bosnia and Herzegovina declared independence.Footnote 133 At that time, it was a multi-ethnic State, made up of mainly Muslim Bosniaks, Orthodox Serbs and Catholic Croats.Footnote 134 Bosnian Serbs objected to this declaration of independence. Supported by the Serbian government of Slobodan Milošević and the Yugoslav People's army, Serb forces attacked the Republic of Bosnia and Herzegovina to unify and secure Serb territory. A struggle for territorial control followed, accompanied by the ethnic cleansing of the non-Serb population from areas under Serb control; in particular, the Bosniak population of Eastern Bosnia, near the border with Serbia.Footnote 135
Many areas in Bosnia were designated as safe zones during the war. This section's focus on Srebrenica thus needs justification. Srebrenica was the first safe zone established in Bosnia, declared about three weeks ahead of the others (Sarajevo, Tuzla, Bihać, Goražde, and Žepa). Particular attention is given to whether there was a real risk of persecution in this section because Srebrenica ultimately fell to the Bosnian Serb forces.Footnote 136
As the armed conflict spread to Bosnia-Herzegovina in 1992, non-admission policies were being implemented by European States.Footnote 137 In July 1992, few countries responded positively to a German call on governments to pledge to accept quotas of refugees.Footnote 138 However, many States supported calls from Austria, Netherlands and Slovenia to establish safe zones in Bosnia.Footnote 139 The International Committee of the Red Cross stated that:
As no third country seems to be ready, even on a provisional basis, to grant asylum to one hundred thousand Bosnian refugees … an original concept must be devised to create protected zones … which are equal to the particular requirements and the sheer scale of the problem.Footnote 140
Thus safe zones were established primarily as an alternative for asylum,Footnote 141 which was explicitly acknowledged by States,Footnote 142 and many asylum claims were refused on the grounds that ‘safe havens’ were available in Bosnia.Footnote 143 Many States justified their refusal to accept refugees on the grounds that they would be contributing to ethnic cleansing, and also relied upon the ‘right to remain’ of those displaced by conflict.Footnote 144
In Resolution 819, the Security Council declared Srebrenica as a safe zone.Footnote 145 However, the declarations of safe areas did not commit the UN to defend these areas militarily. On 4 June 1993, the UN Security Council extended the United Nations Protection Force's (UNPROFOR) mandate by enabling it ‘to deter attacks against the Safe Areas’ and authorising it ‘in self-defence, to take the necessary measures, including the use of force, in reply to bombardments against the safe areas by any of the armed parties or to armed incursion into them’ (emphasis added).Footnote 146 It is important to emphasise here that no enforcement measures were provided for. Rather, UNPROFOR's presence in the safe areas was requested in order to deter any attacks. Barutciski has described it as ‘as a modest endeavour to provide asylum for refugees coupled with ineffective declarations espousing preventive protection which were not backed up by a genuine commitment’.Footnote 147 Moreover, of the 34,000 troops demanded by UNPROFOR, only 7,600 were provided;Footnote 148 the zones were not demilitarised; and the Bosnian army used them as a training ground and for launching attacks.Footnote 149 Thus the Bosnian Serbs perceived the safe havens as helping Bosnian forces to maintain territory. Although the safe havens did save lives and offered some sanctuary to those fleeing violence, history has perceived them to be a failure. Bosnian Serb forces entered Srebrenica in July 1995, following which more than 7,000 Muslims were killed.Footnote 150
There is widespread consensus that the safe zone of Srebrenica was a failure.Footnote 151 In terms of the baseline IPA criteria, Srebrenica did not satisfy the criteria of accessibility. Srebrenica was declared a safe area partly because many individuals had already fled there to seek protection. However, it could not be perceived as an area where individuals could be expected to relocate to because it was surrounded by enemy forces,Footnote 152 clearance for the movement of people into the safe zones had to be obtained from the Bosnian military or government,Footnote 153 and small-scale fighting flared up continually along the outskirts of Srebrenica from mid-1993 until its fall in 1995.Footnote 154 Therefore it could not have satisfied the criteria of accessibility.
It is also very clear that the other two criteria of protection from persecution and no new risk of refoulement were not satisfied by Srebrenica. This is unsurprising because, as aforementioned, commentators generally describe the establishment of this safe zone as a failure. Srebrenica was captured by Serb forces, leading to the deaths of more than 7,000, which clearly amounts to persecution.Footnote 155 The conditions in Srebrenica were reported to be dire which may (but not conclusively) amount to refoulement, as not all violations of human rights will necessarily amount to refoulement.Footnote 156 However, many people fled the zone and were subsequently killed by Serb forces, which amounts to indirect refoulement.Footnote 157
Finally, it is worth noting that Australian courts explicitly found that the safe zones in existence in Bosnia in 1994 and 1995 did not constitute an IPA because ‘those zones are far from being safe and that their inhabitants are extremely vulnerable to “ethnic cleansing” and other persecutory practices by Bosnian Serb forces’.Footnote 158 It is therefore clear that Srebrenica did not satisfy the criteria for a valid IPA.
Some of the safe zones established in the past were deeply problematic, unworkable, and ultimately failed to provide effective protection from persecution. This article posits that refugee protection and diplomatic solutions should be the primary solution to refugee flows. However, it is also important to seriously consider safe zones as a potential response to refugee flows. In situations where granting asylum proves extremely challenging—such as where borders are closed, in situations of mass influx, or where individuals are unable or unwilling to leave their State—safe zones may be the only feasible option to save lives.
Moreover, given safe zones’ long history in international affairs and Turkey's recent establishment of a safe zone in Syria, it is crucial that we understand the legal framework surrounding safe zones. Despite the rise of anti-refugee sentiment and populist narrative worldwide, States are keen to be seen as compliant with their international law obligations. Most States perceive it in their interests to maintain a system of predictable and peaceful co-existence and cooperation. In addition, States may obey international law out of fear of reprisals from the international community, because they feel a sense of morality or solidarity with other States, or perhaps for reputational reasons and to strengthen their international influence. Thus, in the oft-cited words of Louis Henkin, ‘almost all nations observe almost all principles of international law and almost all of their obligations almost all of the time’,Footnote 159 and this proposition has been largely confirmed by empirical scholarship.Footnote 160 Clarifying the legal framework surrounding safe zones can thus guide future State behaviour and prevent safe zones from becoming completely divorced from any legal parameters.
This article has provided the first detailed legal analysis of the refugee law framework surrounding safe zones. It has sought to enrich the IPA debate by clarifying the applicable criteria in the IPA inquiry: (i) the proposed IPA must be accessible; (ii) there must be effective protection from the original risk of persecution; and (iii) there must be no new risk of persecution or of refoulement. It stands apart from other work as these criteria are rooted in State practice. In addition, this article has taken a novel historical approach to show the challenges in invoking safe zones as an IPA such as the difficulties in ensuring the immunity of a safe zone from attack, and the difficulting in establishing an accessible safe zone in the midst of armed conflict. The article also used the example of Operation Provide Comfort to demonstrate that safe zones, in very limited exceptional circumstances, may satisfy the IPA criteria.
The arguments made in this article are primarily of relevance to those involved in the refugee status determination process—namely advocates, judges, relevant international organisations and NGOs, and asylum-seekers themselves. However, the article also serves a broader purpose of challenging the practice of using safe zones as a tool of containment. It is important to note that most of the time, whether a safe zone qualifies as an IPA is not examined by refugee status determination bodies. There is widespread acceptance that one of the primary motivations—if not the primary motivation—for the establishment of safe zones is refugee containment. Most displaced persons never leave their State, and States might invoke safe zones as a ground for closing borders. Thus the issue of whether a particular safe zone qualifies as an IPA never makes it as far as a tribunal or a court. The structure of refugee law, and international law more generally, coupled with the practical realities of human displacement, mean that legal protections in the sense of administrative or judicial determinations of entitlement are only realistically available for the very few people who actually make it to the territory of a foreign State. Therefore, the place that international law can really ‘bite’ is in what becomes generally accepted among States as compliance (or not). By identifying the applicable IPA framework, this article will hopefully inform States’ future plans in respect of establishing safe zones, could help entail that future safe zones have more success than in the past, and will provide a benchmark for those seeking to assess whether a particular safe zone constitutes an IPA.