I am grateful to Matthias Hoesch for his impressive engagement with some of my recent work on refugee protection (the essay on which he focuses is the third of four pieces that respond, at least in part, to Miller’s evolving arguments).Footnote 1 The essay on which Hoesch focuses has three parts. In the first, I offer an account of the international refugee regime that draws attention to two issues: first, the priority of the duty of refugee protection over the fair distribution of refugee protection that is expressed in the principle of non-refoulement and, second, the collective action problem set up by the absence of practices of fair distribution. In the second, I address and reject Miller’s argument concerning fair shares and defend the priority of refugee protection while arguing that we have reason to ensure as far as practicable that distributions are fair. In the third part, I address the problems that attend developing an account of fair distribution of refugee protection and how this may be addressed. Hoesch’s comments are primarily directed at the second part of this paper. In this response I will not take up every point that Hoesch advances but focus on what I take to be the key issue between us.


Perhaps the best way to begin is to sketch the three different positions in play in this debate. The first, which Hoesch and I both reject, is advanced by David Miller and can be sketched thus:

  1. 1.

    States have a duty of justice to do their fair share of refugee protection, where fair shares are specified under the presumption of full compliance.

  2. 2.

    Once they have done their fair share, they may have humanitarian duties to take more if they can do so withot bearing unreasonable costs but they have no further duties of justice.

  3. 3.

    The distinction between duties of justice and humanitarian duties matters since an agent may legitimately be coerced to comply with duties of justice but not with humanitarian duties.

I should note that when in relation to my first argument against Miller (i.e., that given his distinction between duties of justice and humanitarian duties, one might wonder how he can sustain the idea of duties of justice to refugees since it could be argued that each state does its fair share in respect of duties of justice by protecting the human rights of its own population and respecting human rights of others), Hoesch’s endorsing gloss concludes: ‘Thus, they [states that protect their residents human rights and don’t breach the human rights of non-residents] would not have any duty of justice towards those asylum seekers; the whole responsibility for the failure of basic rights protection would be on the shoulders of the rogue states.’ The first part of this is right, but we should slightly qualify the second part since human rights-compliant states would still have responsibilities grounded in humanitarian duties and asylum seekers would still have moral grounds for complaint against them even though the primary responsibility lies with non-compliant states.

The second position is my own expressed in the second argument against Miller and can be summarised thus:

  1. 1.

    States have a collective remedial duty of justice to refugees.

  2. 2.

    In contexts of remedial responsibility, a duty of justice is a duty to act effectively to remedy the injustice and this requirement of effectiveness takes priority over issues of fairness in the final distribution of responsibility among states.

  3. 3.

    To insist that justice is limited to doing one’s fair share (as specified under the presumption of full compliance) is illegitimately to shift the cost of the failure to ensure full compliance, when this occurs, from the group who owes the collective duty and is responsible for ensuring compliance onto the group to whom it is owed. Thus, in cases of collective duties of justice, there is an obligation to accommodate fairness as far as possible within the duty to act effectively but not at the cost of failing to act effectively.

  4. 4.

    Hence the duties of justice of states to refugees may extend beyond doing what would count as fair shares under conditions of full compliance when these conditions do not obtain, that is, a compliant state’s fair share can vary depending on the compliance or non-compliance of other states.

The third position is that of Hoesch and can be specified thus:

  1. 1.

    Every state is obliged to do its fair share of refugee protection.

  2. 2.

    Every state which already fulfills its fair share has the duty to accept, in addition to its fair share, all refugees (i) who are on its territory or at its borders and (ii) who the state in question cannot effectively transfer to a country where the refugee will be safe and protected from non-refoulement that has not fulfilled its obligation to the extent to which the state in question has fulfilled its own.

  3. 3.

    Every state is allowed to use border policies to reduce the number of asylum seekers that reach its territory to its fair share if those border policies do not coerce third states to carry more burden than this state itself does.

How are we to decide between these positions? To address this debate, consider a variant on an example that Hoesch deploys: the world is comprised of a failed state R and 5 functional states with identical capacities (A, B, C, D, E). A is the only neighbor of R and bears 12 units of refugee protection, B, C & D bear 4 units each, and E has a burden of 6 units (exactly its fair share). B, C, D refuse to take anymore and this is common knowledge to A and E as is also the knowledge that they cannot effectively coerce B or C or D to take any more refugees. What do each of the positions have to say about the duties of B, C, D and E in this context?

All three views hold that B, C and D are in breach of their duties of justice and at least Miller and I hold that B, C and D would not be wronged if they were coerced to comply (either by A or E or A and E or a legitimate international institution).Footnote 2 In the absence of effective sanctions capable of altering the conduct of B, C and D, the three views take different paths.

Miller’s view holds that E has no obligation in justice to take any more refugees, however, it acknowledges that compliers do have humanitarian duties to take up the slack given the dire need of the victims of injustice who remain unprotected. Although the primary complaint of these victims must be directed at the non-compliers who bear responsibility in justice for their unprotected condition, they can have a secondary complaint against compliers who do not take up the slack generated by partial compliance if compliers have the capacity to do so without incurring unreasonable costs.

My own view argues that the collective duty on A, B, C, D and E to remedy the injustice suffered by refugees requires that in circumstances in which B, C and D refuse to do their fair share and cannot be effectively coerced into doing so, then A and E as states that respect the obligation to remedy the injustice have an duty to fairly share the responsibility for the resultant additional refugee protection between them, subject to a demandingness limit and in accordance with their relative capacities, which would imply, in this context, E accepting 3 units from A given it is stipulated by the example that A and E have identical remedial capacities. Consider, as an analogy, a neighborhood swimming spot (a river bend, a disused quarry, etc.) where unsupervised children regularly drown or risk drowning and where there is a collective obligation on the neighborhood community to ensure that supervision and rescue capabilities are reliably available. Suppose our neighborhood community required all adults to attend lifesaver swimming classes and to serve as lifeguards on a rota so that five trained adults (with identical capacities) were always present to supervise and rescue at the local swimming spot when required – and it has the legitimate authority and power to impose this rule via sanctions over members of the community but chooses not to do so. In a rescue situation involving thirty drowning children, Alf and Ed find themselves with Bert, Charlie and Doug who each rescue four children and then declare that they have done enough and wander off home. Faced with such partial compliance, on my account, Alf and Ed each have a duty of justice to rescue nine children each (as long as doing so does not unreasonably risk their own lives). The fact that Alf and Ed are part of a community of agents whose collective choices result in the failure to ensure Bert’s, Charlie’s and Doug’s full participation in doing their fair share generates a duty of justice on each of them to do more than what would be their fair share under conditions of full compliance and to divide the excess responsibility fairly between them (I am assuming here, to keep the analogy on a par with the previous example, that Alf and Ed possess identical capacities).

On Hoesch’s view, E having done its fair share has the right to refuse any transfer from A and to adjust its border policy to prevent further claimants arriving. If I understand him correctly, the logic goes like this:

  1. a)

    A and E have both fulfilled their fair share (1) and non-refoulement (2i) obligations.

  2. b)

    Given A is not able to coerce B, C and D to take more burden, and if there will be sanctions to B, C and D that are more severe than the threatened sanctions to E, it is plausible to assume that A might legitimately coerce E. Hoesch characterizes this as A’s right to self-defense against having to bear an unfairly heavy burden.

  3. c)

    However, E would have a right to resist, because E has no duty [of justice, I assume] to accept more refugees.

  4. d)

    Hence, E is not wronged by A if A coerces E, and A is not wronged by E if E successfully resist A’s coercion and does not accept any more burden.

Let us now address these views critically.

It is worth stressing that all three views agree on two points. First, that the best scenario is one in which there is an effective global refugee regime that secures full (or nearly full) compliance of state’s with doing their fair shares (where fair shares are determined of a basis that none can reasonably reject) – and that all states have a natural duty of justice to support the construction of such a global refugee regime. Second, that this is not the situation at hand and is not likely to be in the immediate future. Beyond that initial point of agreement, they diverge. Both Hoesch and I claim to reject Miller’s limitation of duties of justice to doing one’s fair share as it would be specified under conditions of full compliance. In my own case, this rejection is integrally bound up with an account of the relationship between effective protection and fair distribution in which the former takes priority but where the latter should be as fully realized as compatible with delivering effective projection. In rejecting Miller’s view, I both acknowledge the force of the intuition motivating that view, namely, that it seems counter-intuitive to say that what properly counts as an agent’s fair share of responsibility varies according to the degree of compliance or non-compliance of other actors and argue that defeating Miller’s argument requires showing that fair shares of responsibility can vary on this basis in contexts of remedial justice precisely because the constraint of effective discharge of the collective duty has priority, but that this constrains rather than removes the obligation to try and ensure maximal fairness. Hoesch agrees that effective protection should take priority but does not take it to be a requirement of justice that fair distribution be realized as fully as possible within this constraint in contexts of non-compliance. In other words, he wants to preserve Miller’s intuition that one’s fair share of responsibility is what that share would be under conditions of full compliance but argue that one may have duties of justice to take up a greater burden of refugee protection on grounds other than fair distribution – and I take the thought to be that respect for the principle of non-refoulement is such an alternative ground.

At this stage, I need to address what may be a matter of (linguistic?) confusion by either Hoesch or myself. In his commentary, Hoesch argues that neither my first or second argument against Miller justify the claim that the resdistribution of refugees who are unprotected due to non-compliance of some states with doing their fair share should be done fairly (that is, according to their relative capacities). The issue here is that to say that an argument does not justify a claim can have two different senses: an internal sense in which it expresses the judgment that the argument is defective on its own terms or an external sense in which it expresses the judgment that the argument fails not due to internal defects but because it fails to register the possibility of other more plausible arguments that broadly share the same premises but do not lead to the same claim. If we read Hoesch’s claim as a remark about the internal logic of my arguments, this strikes me as right about my first argument against Miller (which does not, and is not intended to, speak to this issue) but simply wrong about the second argument (as the reconstruction of my position and example given above should illustrate), and wrong in a way that underestimates the originality of Hoesch’s own view. Rather I take Hoesch’s talk of justification to be that of the external sense and his claim to be that it is conceptually possible to reject Miller’s version of the fair shares argument without buying into the commitments that my argument imposes, namely, by subordinating the fair shares claim to the non-refoulement principle (hence the importance of his reconstructive work on Miller at the start of his piece). The sense in which my second argument does not justify the view that refugees unprotected due to non-compliance with fair shares is then not a comment about the logical structure of this argument but a claim about there being more options on the table for rejecting Miller’s fair shares claim than my argument acknowledges in ways that are consequential for the justificatory power of my argument. This point matters to how to read and respond to Hoesch’s argument and it is in this second sense that I will address his criticisms and his proposed alternative.

Hoesch offer three different criticisms of my view. The first is that it is too demanding (here Hoesch makes rather a fuss over what was a throwaway rhetorical remark about Germany and Sweden that was directed at criticizing other EU states). Whether that is a reasonable criticism or not depends on the level of theorizing in which one is engaged and while we are both doing non-ideal theory, my sense is that Hoesch is more concerned to be feasible and policy-relevant than I am. I will leave this aside. The second is that it is shaped by a misleading picture of refugee protection in terms of the ethics of rescue and while my argument works within such terms, it ceases to be persuasive when removed from this framing. This then leads to a defence of Hoesch’s own proposal framed in terms of reasons to endorse that moderate position rather than my more demanding view.

What I take to be the pivotal criticism that Hoesch directs at my view as expressed in this paper is that it hangs on modelling the issue of refugee protection in terms of rescue scenarios. Hoesch accepts that if we consider the standard one-off rescue scenario, then my argument is persuasive – but denies that we should accept the analogy. I am sympathetic to this critical point and argued in my critique of Miller that we should be skeptical of treating the issue as analogous to one-off rescue scenarios, hence the turn to the more continuous rescue scenario discussed above. But perhaps my argument remains too bound up in the rescue picture? Hoesch argues that there are salient differences: given general respect for the principle of non-refoulement, (a) there is always an initial unfair distribution in place and (b) state border policies aimed at preventing refugees reach their shores are never fully effective (so it is not like just refusing to save someone) hence (c) ‘the initial distribution constantly is in flow, depending on decisions of refugees to leave some countries and to seek asylum in others’. These are all true empirical claims. On the basis of these points, Hoesch offers the following claim:

… the rescuer model seems not to be helpful. There are no states that once and for all declare that they will not accept some more burden as they already have carried. In consequence, there also is no determined group of states that are willing to accept more than their fair share in order to take up the slack, and there is no sense in claiming that this group should distribute the burden justly.

The claim seems to be this: because we can’t empirically identify states as complier/non-compliers in a binary fashion, there is ‘no sense’ in even talking about a group that should distribute the burden of unprotected refugees fairly. I would accept that my argument involves a simplifying idealization in working out its view of duties of justice and, hence, that the application of this view is likely to be more complex than the cases discussed in the theoretical picture – but I don’t see that this implies that there is ‘no sense’ in articulating a duty to distribute unprotected refugees fairly according to the relative capacities of state to protect them. If Hoesch means to reject the use of idealization in political philosophy that is a view that would require much more defence than offered here.

Hoesch could, I think, have pressed further on the rescue picture by noting that my community swimming example (see above) supposed that the community had a meaningful choice about whether to have a coercive regime that generates (close to) full compliance or whether to have a less coercive regime. My point was that the same agents who have the duty of rescue are also the agents who are responsible for the scheme of rescue and the options for non-compliance within it so cannot disavow the collective duty of rescue even if some individuals fail to comply. Here Hoesch might say that although states have the collective duty to protect refugees and are collectively responsible for the scheme of protection and how severe or lax the sanctions for non-compliance are within it, they cannot be seen as equivalent to a community making a collective choice because the conditions of their being such a community are not present. This would certainly pose a problem for my argument that states can be coerced to do more than what would be their fair share under conditions of full compliance. I take it that Hoesch may be pointing in this direction when he remarks:

We can … ask if international institutions may coerce E to accept more burden from A. I would deny this. The international order is established by the collective of all states. The majority of those states do not contribute their fair share. Why should the collective, substantially constituted by those selfish states, have a right to coerce E to unburden A? E would have a right to resist.

This strikes me as a reasonable point to make given current conditions although it does not, I think, affect the point that E has a duty of justice to relieve A of some of its excess burden. On that point Hoesch seems on much weaker grounds. He argues:

If state E voluntarily accepted to take some burden from A, then the inequality between A and E would be diminished. However, at the same time, another inequality would increase, namely the inequality between B, C and D on the one hand and E on the other hand.

Yes but, at the same time, the inequality between A and B, C and D would decrease by the same amount, and overall inequality would decrease since 9,9,4,4,4 is a less unequal overall distribution that 12,6,4,4,4. Saying, as Hoesch does, that from the standpoint of an ideal distribution of 6,6,6,6,6 there is no difference between 9,9,4,4,4 and 12,6,4,4, 4 ignores the point that the increase in civic burdensomeness with each additional unit of refugee protection is typically not uniform and linear but rises with each additional unit. A public health service that is comfortable at 6 units may be stretched by 9 units such that each unit added after that requires additional resources compared to the unit cost in moving from 6 to 9 to compensate for the strains on the sytem. So if we accept, as Hoesch claims to, that ‘more justice in distribution is better than less’ and this ‘is why states have an obligation to engage in the implementation of fair distribution schemes’, then it is unclear why we would resist the view that there are reasons in justice for E to accept refugee transfers from A.

What though of Hoesch’s own preferred view? If I have understood it correctly, the key points can be summarized thus:

  1. 1.

    The duty of non-refoulement has priority.

  2. 2.

    States should do their fair share as specified under the assumption of full-compliance.

  3. 3.

    States that have done their fair share can legitimately use non-arrival and other measures to prevent refugees reaching their territory in order to claim asylum (and hence be subect to the non-refoulement principle) as long as this does not coerce any other state to take on a greater burden than they are bearing themselves.

I’ll limit myself to two critical remarks on this view.

The first is simply that it sets up a structural inequality between states. States who share borders with refugee-producing states and cannot thereby avail themselves of non-arrival measures are liable to be heavily over-burdened unless all states do their fair share under conditions of full compliance or some non-proximate states do more than their fair share. In contexts of partial compliance, non-proximate states that do what would count as their fair share assuming full compliance can, on this account, claim the right to refuse to accept resettlement transfers from these over-burdened proximate states and to obstruct refugees seeking to travel to their territories through non-arrival measures. In other words, Hoesch’s argument constructs a structural inequality between the obligations of states on the basis of whether or not they share a border with refugee-producing states. It might be argued that this is acceptable in a world in which it was a matter of random chance which states happen to border refugee-producing states and, hence, this was simply a matter of misfortune. However, this is not our world. On the contrary, which states are liable to be or to border refugee-producing states and which states are distant to those states is, in significant measure, a product of wider global injustices both historic and current. To insist that, under our non-ideal conditions, it is acceptable as a matter of duties of justice for such already disadvantaged states to have to bear a disproportionately heavy burden of refugee protection strikes me as simply adding insult to injury.

The second criticism is that Hoesch’s argument sets up perverse incentives. On his view, a disproportionately burdened state that is unable to relieve itself of a reasonable measure of its burden thereby has both legitimate reasons and prudential incentives to try and support refugees in bypassing of the non-arrival measures deployed by other states, that is, to aid and abet people smuggling. Because Hoesch’s view combines the priority of the non-refoulement rule with both a rejection of duties of fair distribution according to relative capacity of unprotected refugees under conditions of partial compliance and an endorsement of legitimate coercion (as ‘right to resistance’), it thereby allows an over-burdened state engaging in the subversion of a less burdened state’s border practices in order to facilitate refugees gaining presence on the other state’s territory and, thereby, becoming subject to the over-riding norm of non-refoulement. Legitimating such a dynamic is not, I submit, likely to be prudent route to adopt.


I have not touched on every point or disagreement raised by Hoesch’s interesting arguments (or in clarifying every point where I disagree with his representation of my view), I hope readers will return to the essay on which Hoesch focuses to evaluate both his critique and my response. I am though grateful to him for raising a number of important issues and forcing me to at least partially rethink my position.