Restructured Safe Havens: A Proposal for Reform of the Refugee Protection System

Ahilan T. Arulanantham

I. Introduction

This project arose from questions that I asked myself subsequent to a disturbing summer's research over one year ago. While in south India this past summer, I had the opportunity to speak with Tamil refugees who had fled Sri Lanka during the previous few months. I was deeply dismayed to find that many of them had fled a United Nations High Commissioner for Refugees (UNHCR) camp designed to provide safety within Sri Lanka for the persecuted Tamil population. ¹ Forces of the Sri Lankan Government's army had begun to arbitrarily take people from the UNHCR's camp. Many of those had returned with tales of torture, and still others had not returned at all.

Though these accounts were disturbing, the UNHCR's reporting on the situation was even more troubling. Refugees recounted to me how UNHCR officials attempted to help them recover disappeared friends and relatives (with mixed success) and that at a certain point, the camp's director actually [End Page 1] told the refugees that the UNHCR could no longer guarantee their safety. ² Yet, at a time when disappearances were increasing and thousands of refugees were fleeing the camp, ³ the UNHCR published a report claiming that the camp was open and providing protection and relief for those seeking shelter from the conflict. The report made no mention of the fact that protection failures in the camp were rampant and that many refugees were risking their lives on the high seas in order to leave it. ⁴

To summarize, I had come upon a situation in which the UNHCR had created a camp designed to protect internally displaced people within their home state. People did not leave Sri Lanka (and, in some cases, returned to Sri Lanka from India) in part because they believed the UNHCR would provide protection. ⁵ When that protection failed, however, the UNHCR was not willing to publicly report it in a way that would ensure the recognition of those refugees' asylum claims abroad, even while UNHCR officials willingly acknowledged that failure to the refugees themselves. ⁶ Why did the UNHCR fail to report such a horrible situation?

While it is tempting to demonize the UNHCR for its seemingly unconscionable behavior, a helpful answer to this question requires an understanding of the UNHCR's motivations. An analysis of these motivations quickly leads one to some of the most fundamental problems in refugee law. Sri Lanka is paradigmatic of the type of situation that strains the refugee system to the breaking point. [End Page 2]

In Sri Lanka, the UNHCR was attempting to manage a situation of mass refugee movement in which thousands of people could attempt at any time to cross international borders in order to flee violence and political upheaval. The architects of the international refugee system did not design the UNHCR to operate in situations of mass influx. The UNHCR and the international refugee system as a whole were created by the Western bloc nations in 1951 shortly after the start of the Cold War. ⁷ The framers of this system envisioned individuals arriving in small numbers from Communist countries under conditions where granting refugee status to these individuals would prove, for the most part, politically expedient. ⁸ While international lawmakers expanded the scope of this protection in 1967, ⁹ they never contemplated the kind of refugee "problem" that I found in India. Today, the paradigmatic mass influx situation is one in which states find it politically unacceptable to take hundreds or thousands (and sometimes hundreds of thousands) of refugees from a country that they do not particularly want to condemn.

Upon returning from India, I learned that the story of the UNHCR's actions in Sri Lanka was not unique. Governments had attacked refugees residing in safe havens created by the UNHCR in Bosnia, Rwanda, Iraq, and elsewhere, ¹⁰ even while other countries refused to give those refugees protection, in part because of the existence of the safe havens. This led me to believe that the situation that I had seen was indicative of systemic problems in the refugee protection regime. If the absence of accurate reporting and adequate protection in safe havens was a recurring condition, perhaps the forces responsible for such reporting and protection found it politically expedient to shirk their responsibilities on a routine basis. [End Page 3]

Not surprisingly, others before me have recognized these flaws. Major proposals for radical change within the refugee system have begun to surface. ¹¹ While there are many, I focus on proposals that aim to deal with refugee protection in situations of mass influx, which in practice requires extending protection to people who face persecution but have been unable to cross an international border. I will organize my discussion around two related (and relatively obvious) problems. First, the refugee system does not adequately address situations of mass influx. While the system sometimes succeeds in dealing with individual claimants, when large numbers of people flee persecution, it simply breaks down. Second, the refugee system affords substantially more protection to people who have crossed an international border than to those who have not. While this distinction makes sense in some cases, in others, particularly in situations of mass influx as a result of persecution by the state, it creates a large gap in the international protection system. I will consider some of the leading proposals for reform that attempt to deal with mass influx situations. After a critique of these proposals, I suggest a different system that I believe offers the best hope of dealing with these inevitably tragic situations.

Three principles guide my analysis throughout. First, and most importantly, I require, as a methodological constraint, that any proposal be politically viable. Any inquiry into the refugee system must take into account that the system is designed by states to protect people who are not their own nationals. In that context, that states provide for the protection of refugees at all is quite remarkable. While one can conceive of utopian systems of ideal refugee protection, that the system is one constructed by states, and therefore, for states, imposes substantial restrictions on any realistic proposal for refugee reform. I take this political viability constraint as a starting point. ¹² Therefore, I impose a requirement that any reform proposal must give states a self-interested reason to comply with that proposal. [End Page 4]

The political viability principle would seem to make any refugee system impossible, assuming that states are only motivated by their narrowly defined self-interest. However, for reasons that I discuss below, states have agreed to abide by a norm that requires, at the least, that they not return a refugee to his or her country of origin after he or she has arrived in their territory. ¹³ While that norm is under attack, and perhaps not operating at all in some contexts, most states will not willingly and openly violate it. ¹⁴ The existence of this norm provides an important tool for constructing an alternative refugee system because it shows that states do not always act exclusively in their narrow self-interest. If they did, they would simply refuse to take nearly all refugees. However, the fact that, in most cases, states refuse to return refugees to their home countries, and that when they do return refugees they do so in covert ways rather than openly admitting it while denying the practice, ¹⁵ shows that the norm of refugee protection can be effective in a politically viable system of refugee protection.

I take the existence of this norm, even in its currently weakened state, as an indispensable asset for use in constructing an alternative system. Reformers must design proposals in a form that does not jeopardize this central protective provision in the current system. The acceptance of this norm generates the second principle: at times, normative constraints can force states to act in ways contrary to their narrow self-interests. ¹⁶ Though no power prevents states from simply deporting refugees, states almost never do so. ¹⁷ Any alternative refugee system should be designed to force states into positions where those constraints can operate. [End Page 5]

The third principle is more substantive, having to do with the way I view refugee problems rather than with the method of my project. I believe that too often in writing about refugee problems, the primary causes of refugee flight are obscured by complex and detailed discussions of what happens to refugees after they have been forced to leave. This leads commentators to conflate the problems of refugees with the problems of migrants more generally. This error of emphasis is observable in the rhetoric of much of the writing about refugee reform, which often conflates refugees with immigrants and treats refugee problems as of a kind with development problems. ¹⁸

While some may argue that maintaining a brightline distinction between these categories is problematic, I choose to view refugee issues as fundamentally human rights issues and to presume that refugees are people who flee political oppression. Therefore, I intend to exclude from my consideration the more generalized oppression of poverty, environmental degradation, and other forces that drive migration. While I recognize that the causes of flight are often highly complex and cannot be neatly defined into legal or analytical categories that distinguish clearly between persecution and other motivations for flight, I nonetheless take this approach as a guiding principle for two reasons.

First, in most cases of mass influx, refugees are fleeing political persecution, usually persecution by the state. ¹⁹ While some may have more of a preexistent desire to leave than others, perhaps for economic reasons, such motivations have little explanatory power in cases of mass influx [End Page 6] where thousands of people choose to flee with little preparation time. In any conflict, that some people flee for economic reasons and others, on occasion, flee environmental disaster (and famine) should not obscure the fact that the primary cause of mass refugee movement is oppression by governments engaged in rampant human rights violations. Therefore, any system that seriously attempts to control the "refugee problem" must also recognize that it is not refugees who create the problem.

Second, much of the writing about refugee flows in the West describes economic migrants arriving under false claims of political persecution and generalized abuse of the asylum system. ²⁰ While some such abuse undoubtedly occurs, it is impossible to accurately gauge the extent of it. The passage of time, drastic limitations on access to accurate information from war situations, and the difficulty of documenting one's own history prior to sudden and unexpected flight make objective determinations at asylum hearings impossible. In the face of these limits, the most skeptical, defensible view is to remain agnostic about the number of people who are not "real" refugees.

At the same time, it is very tempting to assume that the number of "real" refugees is conveniently smaller than what it appears to be. Quite apart from the psychological tendency to disbelieve stories of horrific abuse, ²¹ it is far simpler politically to decide that not so many people deserve protection. Unfortunately, some very recent history teaches us that many people denied refugee status have had strong claims, even though it was politically convenient to characterize their claims as invalid. ²² Absent information suggesting that refugees usually do not face great threats to their personal safety, it is prudent to treat their problems as human rights problems. Refugees choose abrupt long-term dislocation, typically leaving friends, families, and other social support networks in order to go to alien lands with strange customs and legal systems that are often based on different languages. These are circumstances that, presumptively, most people would not choose.

The third principle has two implications for how this inquiry will proceed. First, any serious reform proposal must at least consider the effect [End Page 7] that its system will have on governments that cause refugee outflow and how it might generate pressure to decrease the kinds of large-scale human rights violations that usually cause mass refugee movement. Refugee issues are a subset of human rights issues and should be thought of as such. Second, it is, for the most part, an error to think of refugee problems as simply part of larger, deeply intractable problems associated with global wealth inequalities and inadequate development models that contribute to migration more generally. While those problems exist and do contribute to migration to some degree, the movements discussed here are produced by political repression of various kinds, and therefore, should be seen as problems that can be substantially ameliorated without fundamental changes in the world economic order or development paradigm. Attempting to understand refugee crises as part of migration patterns more generally will fail to capture the uniqueness of refugees' plight.

To summarize, I aim to suggest reforms to the refugee system based on the following criteria. Reforms must give states a self-interested reason to participate in the system. In doing so, reforms must utilize and, where possible, strengthen the existing norm against non-refoulement. Furthermore, no system can be evaluated without considering how it will effect the incentives of refugee-receiving states to pressure states that commit widespread human rights violations.

With these principles in mind, I present an alternative refugee system that, if properly implemented, could provide superior refugee protection in cases of mass influx and increase pressure on refugee-producing states to halt political repression, while at the same time providing other states a reason to create this system. The system involves a new model for implementing and protecting safe havens whenever they are necessary. Support for the system is generated by restructuring the information network governing reporting on safe havens, which I believe represents the key to changing the entire structure. In its simplest form, the restructuring involves creating a separate organization whose only function is to report on refugee protection issues during situations of massive displacement. UNHCR will no longer be responsible for reporting on its own protection efforts during mass influx situations (including, most importantly, protection issues in safe havens).

I begin with a discussion of the current system's strengths and weaknesses. The system's greatest strength is the Refugee Convention, ²³ which 157 states have ratified. The Convention requires, among other things, that states not send refugees back to the place from which they fled provided that they are within that state or at its borders. ²⁴ In presenting criticisms, it is [End Page 8] important not to lose sight of the basic strength of the system and the fact that it provides adequate protection for hundreds of thousands of refugees each year. ²⁵ However, the Convention regime suffers from two major flaws: first, it cannot be effectively implemented in the face of mass influxes of refugees, and second, it only protects people who cross an international border in spite of the fact that many people in need of international protection, whom we might otherwise think of as refugees, ²⁶ are displaced within their state's territory.

Against this backdrop, I conceive of my proposal as, at a minimum, a gap-filling measure meant to improve the system in areas where it is not working, while preserving its core. Preserving this core ensures that the norm against refoulement will survive my reform proposal. However, my proposal could function as more than a gap-filling measure. A perfectly functioning safe haven system would work better than existing forms of refugee protection. Such a system would save refugees from the trauma associated with international flight while providing all necessary refugee protection. The hardship of flight should not be underestimated. Displacement entails substantial trauma for refugees. They are forced to leave familiar surroundings and social support networks and learn different languages in order to live in alien societies that have different cultural norms. ²⁷ However, the history of safe havens in the current system, in particular their dubious record of protection failures, justifies caution. [End Page 9] Nothing would be more tragic than for reformers to trade away the system's greatest strength--the norm against refoulement--in exchange for a safe haven system with inadequate enforcement provisions. On the other hand, if the safe haven system were properly functioning, most refugees themselves would choose to remain in the haven rather than face the prospects of fleeing abroad. Over time, fewer and fewer refugees in situations of mass influx would make use of the old refugee structure because a preferable alternative would exist for them nearer to home.

My proposal aims to create this preferable alternative by structuring it to be in the interests of both states and refugees. However, there need be no formal dismantling of the legal structure associated with the current system for such a transition to take place. On the contrary, the continued maintenance of the current system provides an important safeguard in case the implementation of safe havens turns out to be less than a success.

After describing the current system, I evaluate two noteworthy approaches for dealing with the problems associated with mass influx. First, I consider safe havens as currently employed. Although safe havens do exist in the status quo, I believe they are best understood as an exception to the general refugee system, which have developed in an ad hoc manner because more conventional refugee protection by states was politically unacceptable. Therefore, safe havens have developed primarily as a substitute for refugee protection, ²⁸ whereby the United Nations (often the UNHCR), acting with a coalition of states, decides to protect potential refugees in a particular part of the home country so that other states will not have to accept those people as refugees when they flee. ²⁹ I evaluate the success of safe havens as a method of dealing with mass influx and internally displaced people, and find, as I did in Sri Lanka, that while safe havens offer the best solution at a theoretical level, in the current system they are rarely, in practice, an adequate substitute for good refugee protection.

Second, I consider two prominent proposals for reform--the plans proposed by James Hathaway and Peter Schuck. ³⁰ These "burden-sharing" proposals share many features, including a broad commitment to repatriation and the division of refugee obligations among states in a manner fundamentally different from that used in the status quo. In each case, the proposed plan would radically restructure the existing refugee system, with the aim of insulating refugee protection from political pressure to end it, [End Page 10] while at the same time increasing the number of people protected by the system. Both articles are extremely important and insightful, particularly insofar as they attempt to deal with the problems of mass influx and internally displaced people while operating under the political viability principle that I described earlier. However, I believe these proposals are deeply flawed, both because they do not ultimately work if we take the political viability constraint seriously and because they cannot deliver the expanded refugee protection that they promise.

After presenting a critique of both the current safe haven strategy and the burden-sharing proposals, I argue for a different avenue of refugee policy reform based on a substantially different strategy of safe haven creation. In the current system, the UNHCR is responsible for both the protection of refugees in its safe havens and for reporting to states about the success of their protection efforts when states consider whether or not to grant asylum to refugees. This leads to a system in which the states that fund the UNHCR's protection activities are also likely to pressure UNHCR to under-report protection failures, because the absence of protection failures in safe havens enables states to deny refugees protection in their country. Therefore, by keeping protection and reporting responsibilities within the same (vulnerable) organization, the current system decreases the likelihood that safe havens will provide real protection. In its crudest form, the fundamental problem is about the separation of powers. The tension between the UNHCR's mandate and its funding structure provides the analytical starting point for my alternative.

Under my proposal, the UNHCR will remain responsible for protection in safe havens but will no longer be responsible for reporting about the success or failure of the safe havens. Instead, information about the conditions in safe havens will be channeled to states via an organization that is not responsible for protection. For purposes of this paper, I call this organization the Secretariat for Refugee Protection Documentation (SRPD), although this function could be filled by any suitable inter-governmental organization with the aid of NGOs that act primarily as reporting agencies (such as Human Rights Watch, Amnesty International, etc.). If accurate information about the lack of protection in safe havens is available to states when making their refugee assessments, they will be less able to ignore their responsibilities in cases of mass influx by simply pointing to the existence of the safe haven. Then states will be faced with the choice of either accepting refugees by the thousands or providing more substantial protection in safe havens. I believe states would be willing to support greater safe haven protection, although either option would be preferable to the status quo.

In the last section, specifically Part C, I present alternative mechanisms for increasing safe haven protection that could be implemented with greater state support. Therefore, my proposal utilizes the constraints already [End Page 11] accepted by states--the desire to avoid increasing refugee obligations and the refusal to openly violate the non-refoulement principle--to push them to support truly safe havens, and then offers some suggestions for how such safer havens could be implemented. I believe that the increased availability of accurate information about protection failures in safe havens would provide the spark for such increased state support. With this support, a system of safe havens could be put in place that offers better protection to more refugees, without their having to go through the trauma of international dislocation.

II. The Current System

The global refugee system, embodied in the Refugee Convention, developed in the shadow of a combative political climate which emerged immediately following World War II. As a product of the West's diplomatic force, the Convention was initially seen, in large part, as a tool for the political embarrassment of the Soviet bloc. ³¹ In fact, the original Convention applied only to European refugees. "Notwithstanding the vigorous objections of several delegates from developing countries faced with responsibility for their own refugee populations, the Eurocentric goal of the Western states was achieved by limiting the scope of mandatory international protection under the Convention to refugees whose flight was prompted by a pre-1951 event within Europe." ³² While this omission was rectified on paper sixteen years later, ³³ the more subtle Eurocentric Cold War biases of the Convention proved harder to eliminate. ³⁴ [End Page 12]

A. Mass Influx

The most pernicious effects of the Cold War model of refugee protection are structural. ³⁵ The framers of the system created after World War II envisioned individualized refugee determinations based on the particular characteristics and situations of individuals applying for refugee status. In fact, when the system was expanded in 1967 to include refugees outside of Europe fleeing for reasons other than those related to World War II, nations had not faced large numbers of asylum applicants. ³⁶ Therefore, the system put into place was not designed to deal with cases where thousands of people are subject to persecution based on their ethnic background.

That the system was not meant to handle such crises can also be deduced from the procedural protections guaranteed by UNHCR's Executive Committee Conclusion No. 8. Some of the UNHCR's recommended basic requirements are:

(iii) There should be a clearly identified authority--wherever possible a single central authority--with responsibility for examining requests for refugee status and taking a decision in the first instance.

(iv) The applicant should be given the necessary facilities, including the services of a competent interpreter, for submitting his case to the authorities concerned. Applicants should also be given the opportunity, of which they should be duly informed, to contact a representative of UNHCR. . . .

(vii) The applicant should be permitted to remain in the country pending a decision on his initial request by the competent authority referred to in paragraph (iii) above, unless it has been established by that authority that his request is clearly abusive. He should also be permitted to remain in the country while an appeal to a higher administrative authority or to the courts is pending. ³⁷

Though there are no established requirements for implementing the refugee determination process, clearly the process imagined by the framers of this [End Page 13] procedural structure could not easily accommodate thousands of people fleeing over the course of a few months or less. ³⁸

Operating under the assumption that refugees were to be few in number and treated primarily as individuals, international actors have generally not been willing to provide the usual array of refugee protections during situations of mass influx. This denial of protection has been accomplished through a number of different means. First, certain doctrinal changes have been adopted to exclude large classes of people, in spite of their questionable compatibility with the core of the Refugee Convention. For example, in some cases states have determined that people fleeing civil wars are not entitled to refugee protection where the violence does not target them specifically, ³⁹ even though the Convention itself creates no specificity requirement. In fact, many victims of civil war are likely to have strong prima facie claims to refugee status, because often such wars are substantially motivated by ethnic divisions. ⁴⁰ Similarly, countries have at times claimed that non-state actors are per se incapable of engaging in persecution ⁴¹ and have often chosen to grant temporary protected status (TPS) rather than full-scale asylum in cases where there were many refugees coming from a single country, despite indications that the violence was unlikely to be short-lived. The UNHCR has often supported these activities. ⁴²

In addition to these doctrinal measures, states have taken more direct policy action to indirectly prevent refugees from gaining shelter during mass influx situations. Many of the Western states have created sophisticated barriers to refugee entry. ⁴³ Other nations have adopted different approaches. [End Page 14] India, for example, cut off benefits and restricted NGO access to its Tamil refugee camps in 1992 in order to encourage refugees living in camps to return to Sri Lanka (and to prevent others from fleeing to India). ⁴⁴

At other times, states have simply decreased the number of claims they consider legitimate when faced with a huge number of asylum applicants. While difficult to document (because such policies could never be explicit), statistical evidence shows that in some cases states essentially operate under a quota system during periods of intense refugee inflow. "When rates of asylum applications soar, rates of approval often plummet. . . . Given the manipulability of the refugee definition, it is impossible to attribute these shifts entirely to alterations in the content of asylum applications." ⁴⁵ Finally, in a few instances, even these pretenses are dropped, and open refoulement in blatant violation of the Convention takes place. ⁴⁶

While the effects of these kinds of policies can be difficult to document (because refugees often attempt to disappear when forced to return to their home country, and governments are especially interested in masking persecution of returned asylum seekers), failures during times of mass influx are likely to be devastating. A recent Amnesty International report indicates that overly restrictive refugee policies have led to the deaths of people returned to their home countries. ⁴⁷ In Sri Lanka, many Tamil refugees, who returned after extreme pressure from the Indian government and false promises of benefits from the Sri Lankan government, remained trapped in transit camps, in some cases for years. Eventually, many of these transit camps became targets for government attacks, forcing refugees to flee the camps and return to India. ⁴⁸

It should be clear then that the refugee protection system breaks down in numerous ways during situations of mass influx. Whatever its other strengths, the system was not designed to deal with huge numbers of refugees, and it has yet to be redesigned to deal with these situations adequately. [End Page 15]

B. Coverage

Besides its general failure to deal with situations of mass influx even where the people fleeing persecution have legal rights to international protection, the current system also suffers from an inadequacy in the scope of legal protection. Like the refugee system's structure, the definitions of the key terms that define the scope of refugee law were also the products of Cold War limitations. ⁴⁹ This produced gaps in the system's coverage that have become painfully apparent in recent years.

First, the Convention has been applied using a fairly limited notion of persecution. As argued above, though the text of the Convention is compatible with both expansive and restrictive operational definitions of persecution, ⁵⁰ in practice it has often been read to exclude people who are victims of large-scale conflict, non-state actors, or in other ways different from the paradigmatic Cold War refugee. ⁵¹ As argued previously, states often use this technique of definitional limitation during situations of mass influx (as in the case of Bosnian refugees), but they are able to turn to it, in part, because of the fundamental malleability of the Convention definition, especially given the way it has been interpreted differently at various times. ⁵²

Second, and most importantly for my purposes, all of the legal protections of the Convention apply to people who have crossed borders. This is the case in spite of the fact that every conflict discussed in the examples above generated large numbers of internally displaced people who were forced from their homes but nonetheless did not cross an international border when seeking refuge. In fact, the refugee regime creates no obligation for states to actively offer asylum protection, which would create a duty for states to aid refugees trying to flee persecution. Instead, no legal protections accrue to a refugee until he or she has reached the border of a potential host state. This means that the Convention does not explicitly prohibit states from creating barriers to entry.

The Convention's key obligatory provisions (Articles 31, 32 and 33) assume a situation in which refugees, possibly by irregular means, have somehow managed to arrive at or in the territory of the contracting State. The obligation [End Page 16] of States to admit asylum seekers is referenced only in a recommendation attached to the Final Act of the Conference adopting the Convention. ⁵³

Of course, many people affected by these conflicts do not have the ability to reach other states. For them, none of the protective measures available within the refugee system accrue because they are classified as internally displaced people. ⁵⁴ This is true in spite of the fact that many estimates report that the number of internally displaced people is actually greater than the number of refugees. ⁵⁵ Apart from the human rights guarantees that apply to all people (and are notoriously under-enforced), the refugee system provides no formal protection for internally displaced people. ⁵⁶

The implications of these definitional coverage problems for any reform strategy should be clear. The current refugee system arbitrarily discriminates between those who are specifically targeted for persecution and those who are targeted as part of more general persecution, even though the danger to life can be the same in both cases. Similarly, the system arbitrarily discriminates between those who cross borders and those who do not, even though both groups may be displaced for identical reasons and face identical threats from their persecutors. While this article does not take a position on the appropriateness of expanding the Convention refugee definition per se, ⁵⁷ any proposal that can eliminate this arbitrary discrimination will represent a significant advance over the current system.

C. Non-Refoulement

Having discussed the two major weaknesses of the current refugee system, this section ends by noting the system's one great strength. In spite of all the [End Page 17] limitations described above, the fact remains that, apart from times of mass influx, states generally accept a binding norm that a refugee who has fled persecution on the basis of race, religion, nationality, political opinion, or membership in a social group will not be returned to his or her home country. Many scholars consider the principle against refoulement to be customary international law, which speaks further to its widespread recognition (even though the practice of states is undoubtedly a better indicator of what international law requires). ⁵⁸ Additionally, while states have continued to use entry barriers to restrict refugee flow, government representatives at times find the social and political costs of maintaining some barriers unacceptable. ⁵⁹

The inclusion of refoulement proscriptions in other international instruments and its expansion in some of those instruments, as well as in UNHCR policy statements, also point to the strength of the norm. ⁶⁰ Finally, even when states do flout the refoulement principle, they rarely do it openly, which is further evidence of the principle's growing international acceptance. For example, when the US government refused to take in Haitians in 1991, they chose to interdict them on the high seas and detain them at Guantanamo Bay rather than simply refusing to let them land, or deporting them upon arrival, despite the fact that the latter options would undoubtedly have been cheaper. ⁶¹ While I do not in any way condone the US policy, that the government took a somewhat intermediate step is significant.

Signs of disaffection among asylum states do not indicate that the Convention is obsolete. Rather, elaborate and disingenuous strategies to avoid the obligation of nonrefoulement testify to the continuing centrality of the duty and to the accomplishment of those who managed to secure obligatory status for it in 1951. Recognizing their obligation, states still do not readily admit to a breach of Article 33. ⁶² [End Page 18]

Therefore, that states are unwilling to simply deny that they have a duty to abide by the refoulement principle speaks to its strength, given that the only sanction they are likely to face for violating the principle is the condemnation of some members of the international community.

The fact that most states abide by the refoulement principle most of the time has two implications for refugee policy reform. First, while engaging in criticisms of the refugee system we must not forget that a refugee facing imminent danger who can escape his or her country and enter a state that gives asylum to qualified applicants can benefit from state protection. ⁶³ This is remarkable in a system in which states rarely provide substantial benefits to nationals of another state, especially when that state is often likely to be displeased by the benefit. The general acceptance of this limited form of altruism can serve as an asset in any proposed system of reform.

Second, binding altruistic norms that are generally accepted are rare in international practice. Any attempt to reform the refugee system must be particularly cautious when tinkering with the proscription against refoulement. Proposals for change must be tailored to guarantee that nothing in them can be used to dilute the strength of the international norm as it currently exists. As Joan Fitzpatrick warns, "The reluctance of the international community to abandon the 1951 foundation reflects not only a sense that the Convention embodies indispensable and enduring values, but also a pragmatic awareness that hoped-for advances might instead dilute standards of protection." ⁶⁴

Before beginning an evaluation of alternatives that attempt to deal with the flaws of the current system, there are some lessons to gain from the description given above. First, the refugee system created by the Convention rarely operates in situations of mass influx. Though states are bound to treat refugees arriving in large numbers in the same way that they treat refugees arriving as individuals, they simply will not do so in the current system. Therefore, any proposal to deal with mass influx must conceive of itself as a gap-filling measure.

Second, simply asking states to fill the gap will not suffice given the current political climate. Though states are willing to stand by the refoulement principle, they are not willing to take on new obligations in the current anti-immigration atmosphere. As Professor Hathaway puts it, "We can no [End Page 19] longer insist on either the routine permanent integration of all refugees, nor expect all governments, whatever their circumstances, simply to receive and provide quality protection to all refugees who arrive at their territory." ⁶⁵ Merely insisting on the application of the Refugee Convention to situations of mass influx will not produce a politically viable alternative. Therefore, a proposal that hopes to gain state cooperation in a venture to provide better solutions to mass influx crises must give states a reason to cooperate in the new scheme.

Finally, we must remember the current system's chief strength--that states generally accept the non-refoulement principle as binding and are unlikely to openly flout it when faced with situations they find undesirable. This strength must be safeguarded and, if possible, used to improve the system rather than traded away for less certain improvements. ⁶⁶

III. Current Reform Proposals Fail

At the outset, purely for the sake of completeness, it should be mentioned that there are theoretically attractive proposals for reform that this article does not consider. Some advocate expanding the refugee definition to increase international protection, ⁶⁷ while others advocate a treaty creating new binding obligations on states in cases of massive displacement. ⁶⁸ Whatever the merits of these and other similar proposals in theory, they are not considered here in detail because they by and large do not meet the political viability constraint imposed at the outset. States have no reason in the abstract to place these greater obligations on themselves. Additionally, the discussion above makes clear that the international system is not even enforcing existing obligations that states have a duty to fulfill. Therefore, proposing new obligations is unlikely to be fruitful. Instead, I am interested in reform proposals that ensure greater enforcement of existing norms or that restructure norms in a way that actually increases refugee protection. I am not interested in merely creating more legal rights for refugees, which they are unlikely to actually receive. [End Page 20]

A. The Current Safe Haven "System"

Safe havens present a politically realistic alternative increasingly used in the current system. Many of the recent massive displacement situations, including Rwanda, Bosnia, Iraq, and Sri Lanka, have seen the creation of safe havens. ⁶⁹ Because these operations have been conducted in an ad hoc manner without a uniform legal mandate, generalization about them is difficult. Indeed, the operations themselves are diverse. For example, the Iraqi safe haven was defended with extensive military power, including air power, while the Sri Lankan safe havens are secured by only a promise. ⁷⁰ They have in common, however, that all are attempts to protect potential refugees within the borders of their country of origin, so as to relieve outside states of having to provide them with refugee status.

In theory, safe havens provide the ideal solution to refugee problems. Refugees are spared the traumatic experience of having to flee their country. In some cases, such as in northern Iraq, many do not even have to leave their homes. Meanwhile, other countries avoid refugee obligations, and the country that produced the refugees is pressured to stop mistreating them due to the presence of international forces within its borders. ⁷¹ It is important to recognize the significance of this solution even though its practical problems will become readily apparent. For most refugees, the effects of dislocation are highly traumatic. People are uprooted from familiar surroundings, isolated from social support networks, and move to places (typically large refugee camps) where they are extremely vulnerable to exploitation. The move also usually requires adjustment to different cultural norms and sometimes requires a new language. A system that could avoid all of these costs but achieve the same protection would clearly be preferable to the refugee system currently in place.

This theoretical vision is reflected in a new legal right--the right to remain--which the UNHCR has increasingly espoused along with the commitment to safe haven creation. This new right is meant to reflect a focus on addressing the root causes of conflicts that produce large numbers of refugees.

While responding to refugee situations in countries of asylum, the [UNHCR] also started focusing activities in countries of origin, seeking to prevent and [End Page 21] contain refugee movements. . . . Invoking the human right to remain in one's country of origin, the [UNHCR] sought to ensure that people were not forced to flee from their homes in the first place. ⁷²

In theory, the articulation of such a right holds great promise, because ensuring people's ability to remain safely in their countries requires an end to many significant human rights violations in those countries.

In practice, however, those implementing safe havens do not even conceive of them as superior systems of refugee protection. Safe havens are not created on a systematic basis, rather they are generally adopted as a second-best alternative when nations are unwilling to provide refugee protection to those fleeing political violence. ⁷³ Given the ad hoc nature of these structures and their dubious motivation, it should not be surprising that they are riddled with problems.

First, and most importantly, protection in these safe havens has not been adequately enforced. Despite the UNHCR's claims to the contrary, the stories of refugees from Sri Lanka described at the outset clearly indicate massive protection failures in the safe havens there. Similar failures have occurred in Iraq and Rwanda. ⁷⁴ The most notorious example, however, comes from Bosnia. As Hathaway describes it:

The Bosnian "safe zones" were never demilitarized as promised by the U.N. Consequently, they were used as launching pads for government raids, logically attracting Serb reprisals. . . . By at least the spring of 1995, it had become clear that the United Nations' "'safe areas' (we)re the most dangerous places in Bosnia." Fighting and atrocities ultimately led to the deaths of thousands of civilians trapped in such "safe areas" as Srebrenica and Zepa. ⁷⁵

That states would under-enforce protection in these circumstances is not surprising given that the safe havens were created out of a motivation to stop refugee flow rather than a motivation to provide superior protection. [End Page 22] Because the protection problems of safe havens have been well documented by others, those stories will not be repeated here. ⁷⁶ However, the significance of this singularly crucial flaw must not be underestimated. ⁷⁷

Second, safe havens have been created in an arbitrary manner. The intervention needed to produce safe havens has not occurred in every case where it was needed. Instead, intervention has been based on politicized criteria. Barry Posen claims that the limiting factor is the strength of the offending party: "In actual practice, deviation from the sovereignty norm has followed a double standard: when the bad guys are weak, such as Iraq in Kurdistan, intervention pops to the top of the agenda; when they are strong, such as Russia in Chechnya, little is said." ⁷⁸ Paul Freedman adds that it also helps if the offending party is considered a "pariah" state:

The Council has intervened in Iraq and Haiti, two countries where the U.S. had a strong interest, but the Council has not intervened in Sudan, Angola, or former Soviet Republics which have all generated massive displacement. Political deference to Moscow and the threat of a Russian veto make it unlikely that the Council will intervene in former Soviet Republics experiencing massive displacement. ⁷⁹

Some might argue that this political interest principle may also explain why no safe haven was proposed in Rwanda until the French-allied Hutu government was fleeing the Rwandan Patriotic Front. ⁸⁰ However one chooses to describe the motives of these governments, it is clear that the potential protection provided by safe havens has not been extended in a consistent manner. There is no general international commitment to intervene when safe havens would be appropriate. This leaves governments interested in the strategy open to the charge that they are engaged in hypocritical and self-interested behavior. As Hathaway states, "There is no evidence to date of an international commitment to intervene against the [End Page 23] root causes of refugee flows, a condition precedent to the exercise of any genuine right to remain." ⁸¹

The third problem generated by the ad hoc approach to safe havens in the current system has to do with the nature of the legal mandate and intended effects of such safe havens. In some cases, safe havens are implemented only after states consent. In other cases, however, they are implemented pursuant to Chapter VII of the United Nations Charter, thereby authorizing the use of force within a state's territory without that state's consent. ⁸² Additionally, the mandate of protecting forces has varied in different cases, producing substantial problems. For example, the UN guards sent to Iraq were theoretically there only to serve a reporting function, while in Bosnia there was clear intent to protect. ⁸³ In Sri Lanka, the safe havens were created without accompanying security personnel of any kind and are euphemistically characterized as "Open Relief Centres," which suggests that they have no protective function at all even though they were authorized in part as a substitute for asylum. ⁸⁴ The UNHCR later claimed that they were to provide a "relatively safe environment," which they certainly did not. ⁸⁵ At other times, when explaining deployment of forces without consent, the Security Council has defended safe haven operations "stressing that the operations were for purely humanitarian purposes." ⁸⁶

Obviously, this kind of legal ambiguity does not create a stable environment for real refugee protection. Refugees and other governments should only rely on safe havens as substitutes for asylum in cases where it is clear that they can function as such. However, the ambiguous nature of the legal basis of any given safe haven makes such reliance dangerous at best. In spite of this, donor states to the UNHCR support safe havens, and these safe havens almost certainly provide a justification for the denial of asylum claims. While this is extremely difficult to document because reasons for the denial of particular asylum claims are not readily available (and in many instances not made public at all), statistics on asylum cases and occasional [End Page 24] policy statements make the causal connection fairly clear. In the aftermath of the genocide in Rwanda, the French government turned away Rwandan asylum seekers based on the protection afforded by the safe haven. ⁸⁷ Similarly, in 1993, after the safe havens had been created in Bosnia, German asylum adjudicators granted fifty-nine asylum requests filed by Bosnians and denied 1,913. ⁸⁸ Apart from the harm caused by safe havens to those seeking asylum, there are also geopolitical costs. States can use their contributions to safe havens as a justification for decreasing refugee commitments, and their support for safe havens serves to deflect political criticism of refugee policy.

Given these substantial structural problems, it is no surprise that Hathaway provides a blanket condemnation of safe havens in the current system.

None of these interventions [Iraq, Bosnia, and Rwanda] gave the at-risk population a meaningful choice between remaining secure in their own homes and seeking asylum. Access to refugee protection abroad had already been denied, leaving the populations at risk stranded inside their own countries. The 'right to remain' was therefore a hollow rationalization offered by powerful states for their clear infringement of the right to seek asylum. There has been no recognition of the crucial difference between a right to remain (voluntarily, in safety and dignity) and no right to flee (when that is the most effective means to escape danger). Would-be refugees may indeed have remained within their own states, but not because they exercised a 'right to remain.' They had no option but to remain. ⁸⁹

While I agree with Hathaway's account, his criticisms do not necessarily require that the idea of safe havens should be abandoned altogether, particularly given their theoretical attractiveness. I will argue below that it does not follow from these examples that safe havens are inherently flawed but that their failure in the current system is a product of their cloudy legal status, inconsistent use, and the insufficient resources provided to defend the havens themselves.

B. Burden-sharing Proposals

Recently, important new proposals for a different kind of refugee reform have been proposed by two leading scholars--James Hathaway and Peter Schuck. Both advocate different forms of what can be called refugee [End Page 25] burden-sharing. While there are important differences between the two proposals, ⁹⁰ they can be treated together because they represent a type of approach that is important as an option for large-scale refugee reform.

Both plans arise out of a sense that policy reform must be designed to fix a refugee protection system that is breaking down on a global scale, particularly (though not only) in the face of mass influx. Both emphasize that traditional host states are growing increasingly hostile to refugee protection. As Schuck states:

I emphasize instead the burdens that the sudden, massive refugee flows that are now endemic impose on states. . . . [T]he emerging states responses to these [increasing refugee] burdens are seriously jeopardizing the viability of any meaningful regime of international human rights protection. Second, any realistic solution to this problem must somehow forestall these responses by easing these burdens in exchange for a set of obligations that states are more willing to accept and implement. ⁹¹

Both proposals attempt to forestall this breakdown in two ways. First, the burden-sharing proposals call for an expansion in the use of temporary protection status as a mechanism to protect refugees while allowing host states to maintain control over (permanent) immigration into their countries. While both authors argue that repatriation must never be coercive, the explicit goal is clearly to encourage refugees to go home sooner than they do in the current system. Hathaway proposes four principles that should guide repatriation policy to this end including: respect for refugee community social structures, the development of skills and resources among refugees, the promotion of linkages with internally displaced communities within the home country, and the promotion of good information about the human rights situation at home. ⁹²

Schuck argues that refugees must have "decent living standards, access to a fair asylum determination process, and genuine non-refoulement," ⁹³ but his emphasis is clearly on the deterrent potential of temporary asylum as opposed to permanent refugee status.

Temporary protection is also a desirable strategy from the perspective of industrialized states' narrow self-interest. It is a way to keep refugees safely (in both senses) in the Third World from which most of them come. . . . If temporary protection is to succeed in deterring migration to their territories, [End Page 26] these states must ensure that the migrants are protected under conditions of detention, isolation, and privation with little hope of gaining legal status, while also providing levels of safety and hygiene demanded by their domestic standards of decency, if not by the vague common-denominator norms of international refugee law. . . . If potential protecting states come to believe that refuge granted on a nominally 'temporary' basis is likely to become permanent, they will be more reluctant to offer it. ⁹⁴

Greater use of temporary protection serves to give states a sense that they may take control of their borders without diminishing refugee protection. By enforcing the requirement that refugees return to their country of origin when the situation normalizes, the burden-sharing proposals aim to allay states' fears that immigrants will use refugee protection as a back-door to gain permanent residence. The effect of this strategy is to completely untie refugee protection from immigration, thereby allowing states to insulate their refugee protection policies from anti-immigration pressure. ⁹⁵

In addition to expanding temporary protection, both proposals involve a new system of formalized cooperation between nations that act as host states. Under this system, states agree in advance to contribute to global refugee protection, either by committing to protect a certain number of people or by agreeing to fund up to a certain level. Both authors envision a system in which, either through market mechanisms or by prior design, comparatively wealthy states end up contributing more in funds while comparatively poorer states end up contributing more through actual protection. As Hathaway describes it:

Under a regime of common responsibility, all members of the interest-convergence group agree in advance to contribute to protect refugees who arrive at the territory of any state member of the group. States will cooperate in a manner akin to participation in an insurance scheme . . . joining with others to make protection feasible throughout the territories of all interest-convergence group member states. The notion of differentiated responsibility recognizes that it is unrealistic to expect all states to make an identical contribution both to receiving refugees and to financing the costs of the protection regime. . . . Each [End Page 27] participating state would contribute by providing temporary protection, receiving refugees whose special needs make temporary protection inappropriate, resettling those refugees who cannot return home at the end of the period of temporary protection, funding the protection system, or through a combination of these roles. ⁹⁶

Like the proposal I advocate, one of the chief aims of cooperative burden-sharing is to encourage better refugee protection during times of mass influx. As Hathaway explains:

A guarantee of shared responsibility would enable states of the South faced with even a mass influx to remain open to refugee arrivals, secure in the knowledge that a speedy decision would be made to allocate protective responsibilities among partner states. Equally important, because the state of arrival would benefit from a binding guarantee of financial support from extraregional members of the interest-convergence group, its own obligations would be limited to what it could reasonably afford. ⁹⁷

Therefore, Hathaway and Schuck envision a world in which more refugees are protected than in the current system because states can provide greater protection in the knowledge that they will receive help from other states. Additionally, countries will be less likely to renege on their protection obligations because they will not perceive refugee protection as facilitating "back-door" immigration. ⁹⁸ In return for these advances, however, refugee protection is more likely to be temporary (for most refugees) than in the current system, and more of it is likely to take place in countries of the "South."

Before critically examining these proposals, it is important to recognize what they accomplish. First, unlike most proposals for refugee reform, burden-sharing systems offer something to states in an attempt to deal with the current political climate. The burden-sharing proposals recognize that the current refugee protection scheme is under attack, their aim is to create a system in which both refugees and governments benefit. As Hathaway explains:

We can no longer insist on . . . permanent integration of all refugees, nor expect all governments . . . simply to receive and provide quality protection to all [End Page 28] refugees who arrive at their territory. The critical right of at-risk people to seek asylum will survive only if the mechanisms of international refugee protection can be reconceived to minimize conflict with the legitimate migration control objectives of states, and dependably and equitably to share responsibilities and burdens. ⁹⁹

Expanded refugee protection in return for nothing is simply not on the table, and they do not pretend that it is.

Second, the burden-sharing proposals share a crucial insight about how to provide states with a reason to engage in refugee protection. Because nearly every state faces the possibility of burdensome refugee obligations at some point in the future, ¹⁰⁰ they all should be inclined to agree to commit to some share of refugee obligations in order to pool their risk in a manner analogous to insurance. As Schuck explains:

[Some states] might be attracted to burden-sharing for the same reason that many individuals are attracted to catastrophic health insurance: States may rationally prefer to incur a small and predictable protection burden now in order to avoid bearing large, sudden, unpredictable, unwanted, and unstoppable refugee inflows in the future. . . . As the world grows smaller and more interconnected . . . such 'refugee crisis insurance' might well be a 'good buy'--perhaps even for relatively insular states. ¹⁰¹

Regardless of what one thinks of temporary protection and the cooperation scheme in general, the pooling of risk in the face of mass influx presents an important tool for securing state support for enhanced refugee protection, whatever form that protection ultimately takes. [End Page 29]

Finally, the burden-sharing proposals attempt to create a balanced allocation of resources for refugee protection, and thereby address the tremendous inequality in the current system. This inequality operates at two levels. First, though far more refugees are protected in the "South" than the "North," ¹⁰² this asymmetry does not result in more international resources being allocated to refugee protection in the Southern countries. On the contrary, refugees living in camps in the "South" often face abject poverty, even as many (though by no means all) refugees who have the resources or good fortune to make it to the "North" can live at relatively high Western standards of living. As Hathaway states:

[T]he duty to protect refugees is primarily allocated on the basis of accidents of geography and the relative ability of governments to control their borders. States closest to countries of origin and those least able to afford systematic border controls or technologies of deterrence will inevitably receive the most refugees. . . . This chaotic distribution of the responsibility to provide refugee protection is not offset by any mechanism to ensure adequate compensation to those governments that take on a disproprortionate share of protective responsibilites. ¹⁰³

Moreover, this inequality in support to governments is mirrored by inequalities in support to refugees. Hathaway continues:

[I]n 1993, UNHCR allocated more funds to refugee protection in Europe alone than it did for the protection of three times as many refugees in Africa, Asia, and the Middle East combined. Even by 1995, UNHCR had spent less to assist the nearly 1.7 million Rwandan refugees in Burundi, Tanzania, and Zaire than on its residual material assistance and other programs inside the former Yugoslavia. ¹⁰⁴

Thus, in the current system 80 percent of the world's refugees live in the "South," ¹⁰⁵ while "Northern" states increasingly close their borders to large groups of refugees who arrive at them.

In theory, burden-sharing can serve to alleviate these inequalities in two ways. First, a formalized cooperative regime would ensure that the developed [End Page 30] states pay compensation to developing states for their refugee expenditures. Under Schuck's system, this structure would not be mandated but would result from market interactions in most cases, while for Hathaway such a system would be more clearly formalized. In either case, developed states would incur a binding legal obligation to support refugee protection in the developing world where most refugees are, whereas, under the current system any money they provide for this purpose is on an informal charitable basis only.

Second, by protecting more refugees in the developing world, where the total cost of living is less, the burden-sharing proposals aim to ensure that approximately the same amount of money is spent on each refugee. This would be an improvement over the current system's great disparity in resource use between refugees protected in the "North" and those protected in the "South." In addition to a fairer distribution, such a system would allow for greater refugee protection. As Hathaway explains, "under a more collaborative approach to refugee protection, the same resources presently spent to receive refugees could be reassigned where they would be most likely to benefit the greatest number of refugees." ¹⁰⁶ In other words, more people could be protected for the same amount of money if they were protected in countries with lower standards of living.

While the burden-sharing proposals provide valuable tools for refugee reform, they are quite problematic as global alternatives. While I have many substantive criticisms, some of which I will discuss in detail below, one broad problem is worth stating at the outset. Both proposals attempt to use temporary protected status (TPS) as a mechanism to solve or at least alleviate Northern states' immigration enforcement problems. For Hathaway, this takes the form of the explicit claim that, with TPS, money spent on deterring illegal entry could be used for refugee protection once Western nations no longer feared that refugee protection systems would be abused to cause illegal entry. Schuck makes the more straightforward claim that countries would no longer refuse to take refugees because they fear that they are really migrants in disguise. In both cases, the use of TPS is motivated by the sense that, under the current system, states take fewer refugees because they believe these refugees are in fact migrants. On this view, what really motivates states to limit refugee flows is their fear that most refugees are illegitimate; states would be happy to take refugees if they only knew that they were not immigrants in disguise. ¹⁰⁷ [End Page 31]

This account of the current system is troublesome. While, undoubtedly, host government officials are inclined to say that applicants they turn away are not genuine refugees, there are several reasons to be cautious about taking such claims at face value. First, the refugee system breaks down most noticeably during situations of mass influx. It is at these times that states are most inclined to flout the refoulement principle or at least deny a very high percentage of asylum applications. However, these are not the times at which false claims are likely to be a substantial problem. On the contrary, when tens of thousands of people all suddenly flee from their homes, this is when one should be most inclined to think that claims are valid. Nevertheless, it is at precisely these times that governments are most inclined to ignore their refugee obligations. If countries were truly concerned primarily about abuse of the system then they would be more, rather than less, inclined to satisfy their obligations during situations of mass influx.

Second, as I explained above, no one can effectively document how many refugee claimants are "actual" refugees. We should be skeptical of officials who claim to be more than agnostic about the validity of most refugee claims. ¹⁰⁸ However, any policy to turn away increasing numbers of refugees almost has to be justified at least in part by the claim that the refugees are not genuine. These reasons alone should give us pause when attributing decreasing refugee protection in the "North" to increasing belief in the invalidity of most asylum claims. There can be no firm foundation for any such belief, and states will be inclined to make that claim regardless of their true motives.

This diagnosis of the motivation for states' current failures has important implications for how we view the burden-sharing proposals. These proposals take substantial risks with the current protection system, particularly through the advocacy of increased temporary protection. The reward in return for this risk is that their system assuages states' fears about illegal immigration disguised as refugee protection. ¹⁰⁹ However, if this fear is not what primarily motivates states to violate refugee norms, the risks to the system may provide little positive return. [End Page 32]

Apart from this general strategic problem, I have several substantive criticisms of the burden-sharing proposals. First, a strong emphasis on the increased use of temporary protection and a focus on the speedy return of refugees makes a certain amount of involuntary or coercive repatriation almost inevitable. While both Schuck and Hathaway emphasize that return must never occur unless the refugee truly believes that return is safe, the states implementing cooperative burden-sharing have little reason to be so careful about allowing only genuinely voluntary repatriation, particularly given that most refugees will be living in areas closer to their home countries under a burden-sharing regime. B.S. Chimni has persuasively argued that the meaning of policies to "encourage," "promote," or "facilitate" return could easily degenerate into involuntary repatriation under the pressure of Western governments that have strong motivations to ignore the qualifications suggested by people like Schuck and Hathaway. ¹¹⁰ He concludes with a critique that strikingly anticipates the arguments that Hathaway makes.

Only when [words like 'facilitation,' 'promotion,' and 'encouragement'] are looked at in the backdrop of the paradigm shift taking place, and the restrictive practices which have all but become the norm in the developed world that they can be given content. In general, the concern today is less with the refugee community, or for that matter with the host countries, which in the case of 90 percent of the world's refugees is the developing world, but with the need to ensure that refugees do not disturb the peace of the developed world, or invite financial allocations which, we are told, they can ill afford. In other words, the developed States will attempt to give these words expansive meaning for it would help them contain the problem in the developing world. While voluntary repatriation is perhaps the ideal solution, its idealness should not be the pretext to coerce refugees to go back to situations from which they fled in the first place. ¹¹¹

Chimni concludes that the political motivations of states make a focus on temporary protection with an emphasis on repatriation a dangerous tool in the current political climate. ¹¹² [End Page 33]

The history of the UNHCR's promotion of repatriation also gives one little reason to hope that the Schuck and Hathaway proposals can be implemented with complete respect for voluntary repatriation. Several different historical accounts suggest that some of the large-scale refugee repatriation movements organized by UNHCR have occurred with at least some amount of coercion. ¹¹³ While the historical record from all of these repatriations is murky, the point is that the line between voluntary repatriation, coercion, and refoulement is not always clear. This lack of clarity makes efforts to encourage states to promote repatriation more likely to lead to coercion of some kind.

Second, the burden-sharing proposals substantially decrease the incentives for host nations to engage in human rights pressure against the regimes which cause mass influx. ¹¹⁴ The effect of burden-sharing will be to decrease the social and economic costs of refugee protection to host states, while at the same time moving most refugees out of developed nations. As Alan Dowty and Gil Loescher argue, "The previous focus on charity--that is, on relief and assistance . . . is not adequate. Charity alone often helps to perpetuate the injustice that caused the refugee flight, since it relieves the sending country of pressure to correct the injustice." ¹¹⁵ Many of these host nations are those most able to engage in human rights pressure, both because of their international influence and because they are often important aid and trade partners with countries that cause massive displacement. Furthermore, as Deborah Anker argues, the admission of refugees into developed nations itself plays an important role in producing more informed human rights policies towards home countries. ¹¹⁶ Refugees provide important information about what is happening in their home countries that often is not available through the media. This information can then be used to more effectively lobby host country governments. While greater foreign pressure is not the panacea for human rights problems, it undoubtedly plays some role and therefore should not be ignored. [End Page 34]

Third, cooperative burden-sharing systems are likely to have the same kinds of enforcement problems that safe havens currently have--for similar reasons. ¹¹⁷ Under the cooperative regimes envisioned by Schuck and Hathaway, developed states will pay developing states to protect refugees for whom the developed states are responsible. ¹¹⁸ For example, assume the United States has entered into a cooperative agreement with the countries of Central America and the Caribbean, under which it is responsible for protecting 100,000 refugees during a situation of mass influx. ¹¹⁹ Under a burden-sharing system, the United States could pledge to send, if necessary, 50,000 of those refugees to Guatemala, in return for which Guatemala would receive financial or some other kind of compensation (trade benefits, etc.). If large numbers of refugees suddenly began to flee Haiti, the United States could then either admit those refugees and then transfer them to Guatemala or perhaps even direct them immediately to Guatemala. ¹²⁰ Guatemala would be legally obligated, based on the prior agreement, to care for them in a manner consistent with international refugee standards, or (as Schuck says) at least in a manner no worse than they are treated under the current system. [End Page 35]

At first glance, one might think that the host state under such a system (Guatemala, in my example) would have very little reason to abide by international standards for refugee care and protection. After all, Guatemala has essentially bought certain benefits from the developed states, ¹²¹ in return for which it has agreed to care for these refugees. After financial arrangements have been concluded, why would Guatemala maintain a high standard of care? The answer, according to both Schuck and Hathaway, is that the states that bought out of their obligations would be inclined to enforce refugee protection standards in the country in which "their refugees" were residing. ¹²² As this issue lies at the heart of any evaluation of refugee burden-sharing, it will be considered in some detail.

Hathaway makes several arguments for why the developed states would be inclined to protect refugees' rights in the places where they are being held. First, he references the fact that in some European countries and in Canada individual refugees might have standing, as individuals, to enforce their rights. While he acknowledges that at present those rights are limited to people physically present within those countries, he thinks that the jurisdiction to bring these suits could be read broadly enough to include refugees suing from another state. ¹²³ Second, he argues that the refugee protection community in the developed states would be able to "police the quality of protection" at these foreign refugee camps, ¹²⁴ because they could use the existence of inadequate conditions to argue against the validity of removal proceedings for refugees being sent there. Finally, he argues that the system will be likely to send refugees to places where they are most likely to receive appropriate protection, based on the criteria that Hathaway himself suggests. ¹²⁵

Unfortunately, these arguments do not take into account the political realism constraint which Hathaway so vigilantly defends elsewhere in his argument. While a legal argument could doubtless be made to establish legal rights enforceable in the United States for refugees kept in Guatemala, the political incentives of all relevant parties make it highly unlikely that the countries will agree to such a system. Once developed states have rid themselves of refugee obligations by payment, and developing nations are hosting those refugees for a profit, it is unclear why either party would have an interest in keeping those refugees adequately cared for and protected. States are far more likely to agree to sign over legal obligations to the new host state at the time of the agreement, thereby allowing the developed state [End Page 36] to wash its hands of the whole problem. While refugee organizations in the developed states could "police" refugee camps, they would be no more able to do so than they are in the current system, in which events like massacres in safe havens often become widely publicized, if at all, long after their occurrence. In fact, such organizations would likely be significantly less able to police refugee conditions abroad than at home because access to camps could be more easily restricted for foreign human rights organizations and because gathering information about such situations would require more resources than in the current system. ¹²⁶ Finally, while a new institution strictly implementing Professor Hathaway's criteria (which includes choosing refugee locations on the basis of greatest protection provided) would undoubtedly present some kind of an advance, it is far more likely that countries adopting such a system for essentially economic reasons would make refugee allocation decisions based on economic criteria. Needless to say, such criteria would not necessarily place protection concerns at the forefront of decision making.

In Schuck's proposal the enforcement problem appears much more clearly, in part because he considers it explicitly. He recognizes that states that agree to protect refugees in exchange for compensation may be inclined to protect those refugees as cheaply as possible, which could mean providing them with fewer resources. ¹²⁷ This is particularly true in a world where poorer countries may be competing with each other to offer developed countries a lower price for their refugee protection obligations. Any such market-based system carries the risk of such a "race to the bottom," where nations seek to sell their refugee obligations to states that provide the cheapest, and potentially worst, care for their refugee populations. Schuck has three general responses to this. First, he says that states that buy out of their obligations "should be under an independent, continuing legal responsibility to see that the states with which they deal also protect those rights." ¹²⁸ Second, he says that payments should be made gradually rather than in a lump-sum, so that developed states could pressure developing states to ensure adequate refugee protection. ¹²⁹ Third, he says that the [End Page 37] UNHCR (or any other appropriate administrative agency) should have greater authority "to monitor and to publicize the treatment of resettled refugees in light of humanitarian standards." ¹³⁰

Schuck's first two responses suffer from the same political realism problems that were described in the context of Hathaway's arguments. States could agree to remain liable for their original refugee protection obligations, but it is difficult to see what reason they would have to do this. The current system does not establish double liability of this kind, why would states agree to it now? How would it be enforced? Similarly, payments over time would certainly increase incentives to protect well, but developed states would have little reason to carefully monitor such protection, since they benefit financially from low-cost protection under Schuck's system. Schuck's third argument, that the UNHCR should be able to monitor conditions carefully, would provide an important check on decreases in the quality of protection, if the UNHCR would accurately report on such situations. However, it is clear from the current state of safe havens that the UNHCR often succumbs to pressure from governments to remain silent about protection failures. Without fundamental changes to the UNHCR itself, simply exhorting it to monitor and report more thoroughly may not produce substantial change.

After rehearsing these arguments, Schuck seems to basically concede that the quality of protection might decrease under his system. However, he argues that the increase in the number of people protected as more states are drawn into the burden-sharing system makes this sacrifice worthwhile. ¹³¹ While this article will not argue with his normative prioritization (he appears to advocate straightforward utilitarianism), Schuck's characterization seems quite optimistic. One might agree to a straight trade, giving away greater quality of protection for more overall numbers of people protected; however, given the possibly serious enforcement problems, decreases in human rights pressure, and the danger that encouraging quick return could spill over into involuntary repatriation, Schuck's solution seems quite risky.

Therefore, while the burden-sharing proposals offer many useful insights, they ultimately fail as alternatives to the current system. While [End Page 38] burden-sharing would undoubtedly decrease the costs of refugee protection to developed states, it is not clear that the benefits gained from this reduction justify the decreases in protection that burden-sharing may bring. Burden-sharing increases the possibility of coercive repatriation and weakened protection for refugees and does nothing to increase pressure on those states that cause refugees to flee. More importantly, the fundamental problem with such proposals is that they do not increase states' incentives to comply with the system. States only have an incentive to implement the proposals in their initial stages; they have no self-interested reason to continue ensuring the safety of refugees once an agreement has been reached about where to keep them at what cost. Rather than creating a system that gives states incentives to ignore their refugee obligations, reformers must attempt to create a system that gives states an incentive to better protect refugees, either in their home country or abroad. I argue for such a system in the last section of this paper.

IV. The Proposal: Truly Safe Havens

A. The Theory

As we have seen, both safe havens and burden-sharing offer important insights into how to deal with mass influx crises, but neither system ultimately works. Safe havens are the ideal solution in theory, as they allow refugees to avoid the trauma of dislocation to another country while still maintaining their protection under international auspices. However, in practice they function as a substitute for asylum, essentially operating as a meager second-best protection alternative when states refuse to admit large numbers of refugees. Additionally, the international community elects to use safe havens based on highly politicized criteria, and states tend to provide support for the creation of the haven but not its actual functioning as a place of refuge. In other words, states support the creation of safe havens because they provide an opportunity to deny refugees' asylum claims, but then have no reason to ensure the security of safe havens because, once the UNHCR or UN military forces have established the haven, its mere existence presumptively discharges states' asylum obligations. Because information about the conditions of safe havens is poor, states can relieve themselves of their refugee obligations without actually ensuring that safe havens are secure for refugees.

The burden-sharing proposals take a different approach, giving states a reason to accept system reform by offering increased temporary protection and collective insurance pooling. These proposals attempt to resolve the fundamental dilemma facing any refugee reform proposal that the proposal [End Page 39] must get self-interested states to contribute to a system that guarantees protection for people who are not their own nationals--by persuading states that they would be better off in a crisis by agreeing at the outset to contribute to a system of cooperative burden-sharing. However, while collective insurance pooling gives states an incentive to create structures for burden-sharing, it gives them little or no interest in ensuring that refugees will be well treated once those structures are in place (that is, after they have "bought out" of their obligation). Similarly, while states may well agree to implement a strict voluntary return regime with temporary protection, they will have little reason to implement the regime rigorously. Therefore, as with existing safe havens, the most likely scenario under the burden-sharing proposals is one of partial enforcement, where refugees' rights are under-protected. States would receive the benefit of cooperative burden-sharing without ensuring that the system actually aids refugees, just as states benefit from safe havens by their mere existence even if the havens are not actually safe.

My proposal attempts to utilize the strengths of these alternatives while avoiding their weaknesses. In theory, safe havens provide the best solution for refugees and neighboring states; in practice, they are implemented in an inconsistent way and are usually not safe at all. Burden-sharing proposals effectively utilize states' desire to pool risk, but they cannot provide states with a reason to continue to protect refugees. I take the safe haven model as the appropriate structure for refugee protection in mass influx situations but attempt to resolve both the selective implementation and information problems. The same collective insurance pooling motives that drive the burden-sharing proposals can provide states with an incentive to fund and implement safe havens in a more consistent manner. At the same time, altering the structure of the safe haven information mechanism can ensure that states will have a strong political motivation to create truly safe havens. Finally, altered enforcement procedures promulgated by a restructured UNHCR, operating with greater legal and military authority, can lead to more effective implementation of safe haven protection.

B. The Mechanism

The proposal begins with revised incentives. Currently, states have an incentive to create safe havens but not to ensure their proper functioning. Given the absence of consistent and accurate information about conditions in safe havens, their mere existence suffices for states to discharge their refugee obligations. If states were able to discharge their refugee obligations only when safe havens were truly safe, they would have an incentive to create more secure safe havens. [End Page 40]

Here, information provides the missing link. In the current system, the UNHCR itself provides information about conditions in safe havens. If UNHCR publicizes early on that a safe haven is suffering from rampant protection failures, states will be unable to discharge their refugee obligations by reference to that safe haven. Therefore, UNHCR consistently faces tremendous pressure from governments to avoid disclosing protection failures. ¹³² This leads the UNHCR leadership to under-report refugee problems in order to assuage the concerns of potential host governments. The plight of the Sri Lankan Tamils after the infamous 1983 pogroms provides but one example.

[In] 1983 . . . Commissioner Poul Hartling, fearing a government backlash, decided not to press for refugee status for the Tamils arriving in Europe, even though his protection officials thought many of the Tamils qualified as refugees under the 1951 Convention. Here--although it was not intended as such--was an invitation for governments to view the Tamils as 'migrants' rather than 'genuine refugees.' ¹³³

This phenomenon may also explain the under-reporting of human rights violations against Sri Lankan Tamils in the UNHCR's safe haven. Control by governments over the UNHCR has increased in recent years, making this explanation even more plausible. ¹³⁴ The UNHCR is in a difficult position to combat government pressure because the states involved in protection tend to be those that heavily fund the UNHCR and because accurate reporting could lead the state in which the safe haven is located to demand the UNHCR's departure. ¹³⁵ In fact, the UNHCR's safe havens and the rhetoric surrounding the right to remain originated in part because states wanted to decrease their asylum obligations.

As the right of refugees to access secure and dignified asylum fell out of favor in both the North and South, states prevailed upon intergovernmental institutions to devise a less intrusive alternative to the duty to receive refugees. UNHCR responded by proclaiming the 'right to remain.' Given the determination of the governments that control and fund it to deemphasize the right to seek asylum, UNHCR saved face and institutional stature by redefining its raison d'etre, away [End Page 41] from the goal of ensuring access to quality asylum, in favor of an avowed commitment to eradicate the need to flee in the first place. ¹³⁶

Given that states have funded the UNHCR in order to avoid having to face a large refugee influx, it is extremely unlikely that the UNHCR will be entirely forthright in reporting protection failures in safe havens, because the existence of such failures will require the UNHCR's funders to accept the unprotected refugees.

Therefore, my proposal begins by altering the information system. I propose to delink the UNHCR's protection and reporting functions, so that the UNHCR retains protection responsibilities for refugees within the safe haven, ¹³⁷ while other organizations undertake responsibility for providing information about the security situation within the safe haven. Ironically, the Refugee Convention already contains provisions for substantially improved reporting, though these provisions were never implemented. As Hathaway explains, the "UNHCR has never drawn on the express undertaking of Article 35 to formalize a system of periodic reporting and collective scrutiny of respect for refugee rights." ¹³⁸ As explained above, this is not surprising, at least in the context of safe havens, given that the UNHCR's funding comes from states who fear such forthright reporting because they do not want to admit large numbers of refugees.

The reporting responsibilities never adequately fulfilled by the UNHCR should be turned over to a different organization. The organization must be relatively immune from political pressure, and it must also have some expertise in reporting. While all organizations are politicized to some extent, an organization that has relatively low