Using Aid to Control Migration

Abstract

This article examines the practice of using aid to control migration, which we refer to as ‘inducement aid’. We examine two potential objections to inducement aid, each of which concerns a message that the practice communicates to two corresponding audiences: would-be migrants and other developed states. We suggest that the first objection has intuitive force but is undermined by a powerful reply. This finding seems to bolster the intuitive appeal that inducement aid might exhibit as a non-compulsory and apparently option-enhancing form of migration control. However, we argue that the second objection, which targets inducement aid in its capacity as a form of development assistance, has greater power. Developing the second objection, we argue that inducement aid threatens the establishment and maintenance of important international norms, thereby risking degrading the options of the world’s poorest people and setting back the cause of cosmopolitan morality.

‘It ain’t the money, it’s the message’,

– Russel ‘Stringer’ Bell, The Wire

Introduction

In recent years, wealthy states have directed significant quantities of aid to countries from which large numbers of people emigrate. The express purpose of much of this aid is to induce would-be migrants to refrain from leaving their homes and travelling to donor states. This form of aid lacks an official or customary name, but we shall refer to it as inducement aid.

Significant funds have been devoted to inducement aid by both Europe and the US. Since the European migrant and refugee emergencies, European Union (EU) governments have allocated €4.4 billion in aid to a new trust fund aimed at addressing the ‘root causes’ of migration (EUTF, 2020). ¹ The Biden administration considers foreign aid to be a critical element in its strategy to limit migration from the Northern Triangle countries and has called for US$4 billion in development assistance for this purpose (Hesson and Spetalnick, 2021). Experts predict that such initiatives are likely to shift ‘from exception to rule’ (Kipp, 2018).

Inducement aid is Janus-faced. Viewed from one angle, it is a form of development assistance; from another, it is a form of migration control. ² In practice, this distinction is not a sharp one, but it can nevertheless serve to organize our thinking. In its capacity as a form of development assistance, inducement aid is distinguished neither by its immediate aim nor by how it achieves that aim, but rather by what we might call its ultimate aim. Inducement aid’s immediate aim is shared with other forms of development assistance; it is to alleviate some of the hardships faced by some disadvantaged individuals in foreign countries. Inducement aid’s ultimate aim is to reduce migration to the donor state: its immediate aim is expressly presented and understood as a means to this end.

In its capacity as a form of migration control, inducement aid is distinguished from more traditional alternatives by its apparently non-compulsory character. Traditional modes of migration control have a preventive or coercive nature (Miller, 2009). When a state denies visas, enforces its border, and interdicts migrants in transit, it engages in preventive activities. These activities are preventive in the sense that they force certain individuals to refrain from performing a specific act, namely, entering the state in question. When a state deports individuals who enter or remain in its territory without permission, it engages in coercion. Its behaviour is coercive in the sense that it forces certain individuals into performing a specific act, namely, exiting the state in question (Miller, 2009: 114). A common feature of both prevention and coercion is the reduction of options. Prevention removes or constrains the possibility of pursuing one specific (set of) options(s), whereas coercion removes or constrains the possibility of pursuing anything but one specific (set of) options(s). ³

A putatively distinguishing feature of inducement aid is that it is neither preventive nor coercive. On its face, inducement aid does not reduce options, but rather expands them or improves their quality. It does not forcibly exclude or expel, but instead seeks to provide would-be migrants with additional or higher-quality opportunities that can be pursued in their country of origin. The hope of donor states is that, when presented with a suitably expanded or improved option set, many individuals will not need to be forced to stay away, for they will simply choose to do so.

The contrast drawn above between inducement aid, on the one hand, and compulsory forms of migration control, on the other, presents the former in a favourable light. Perhaps because of its apparent innocence, inducement aid has escaped scrutiny from normative political theorists. Our aim in this article is to rectify that oversight.

As the preceding comments suggest, inducement aid must be evaluated on two fronts: in its capacity as a form of migration control and in its capacity as a form of development assistance. For present purposes, that means asking whether inducement aid (a) treats would-be migrants in a morally acceptable fashion and (b) is consistent with donor states’ moral duties to alleviate global poverty. We approach these two questions from a single perspective, examining the expressive character of inducement aid – which relates to the messages it communicates and the potential effects of those messages – and exploring the implications of that character for the treatment of migrants and the alleviation of global poverty.

The upshot of our analysis casts doubt on the claim that inducement aid does not constrain options. Our discussion reveals that although inducement aid may not constrain the options of migrants (qua migrants), it may nevertheless constrain the options of the global poor, and in a way that sets back the cause of cosmopolitan morality. Our analysis thereby exposes one way in which any virtues exhibited by inducement aid in its capacity as a form of migration control can be offset by its shortcomings as a form of development assistance.

We examine two potential objections to inducement aid, each of which concerns a message that the practice might communicate to two corresponding audiences: would-be migrants and other (potential) donor states. We suggest that the first objection has intuitive force but that it is undermined by a powerful reply. This finding might be thought to bolster the intuitive appeal that inducement aid might exhibit in its capacity as a non-compulsory and apparently option-enhancing form of migration control. However, we argue that the second objection, which targets inducement aid in its capacity as a form of development assistance, has more power.

Developing the second objection, we argue that inducement aid threatens the establishment and maintenance of important norms needed to adequately protect the interests of the global poor. It does this by communicating a controversial message about the purpose of aid. By sending this message, inducement aid risks undermining the kind of cosmopolitan ethos needed to protect the basic interests – and enhance the options – of the world’s poorest people.

Before turning to the article’s substantive arguments, two preliminary sections provide a brief descriptive overview of inducement aid (henceforth, IA) and foreground several features of our approach.

The Use of Aid to Control Migration: A Brief Descriptive Overview

The idea that reducing poverty in origin countries can curtail migration is not new. ⁴ However, recent migration emergencies in the US and Europe have prompted governments to increase aid spending with the explicit aim of achieving this outcome. In response to an unexpected surge in migration from Central America in 2014, the Obama Administration created a new, whole-of-government, Strategy for Engagement in Central America (SECA) that aimed, in part, to promote prosperity within the region. The US Congress allocated US$2.6 billion to SECA (Meyer, 2019), and the Biden administration has given the ‘root causes’ approach renewed force. Biden has called for US$4 billion in development aid to Central America and requested US$861 million from Congress for this purpose in his first annual budget proposal (Hesson and Spetalnick, 2021).

In Europe, the use of aid to control migration became particularly prominent with the advent of the migrant and refugee emergency. At the peak of the emergency, in 2015, EU leaders agreed on a new European financial instrument, the EU Emergency Trust Fund for Africa (EUTF, 2020). Since the emergency, EU governments have allocated €4.4 billion in aid towards 224 projects in the Horn of Africa, North Africa, and Sahel/Lake Chad (EUTF, 2020).

These US and European programmes have a variety of different components, including efforts to prevent and fight irregular migration, migrant smuggling, and human trafficking, as well as the repatriation and reintegration of failed asylum seekers and deported migrants. In this article, we are interested only in those devoted to reducing immigration through development assistance. Following Clemens and Postel (2018), we define migration-related development assistance to include economic development, the provision of basic services (shelter, water and sanitation, education, etc.), agricultural support, and projects to improve resilience to environmental shocks. These features have been prominent in both EU and US programming (see EUTF, 2020; The White House, 2021).

Preliminaries

Before we begin our normative analysis of IA, let us make several prefatory remarks about the nature of our approach.

• 1. We shall begin with some comments regarding our moral evaluation of IA as a form of migration control. First, in asking whether IA is a morally acceptable means of controlling migration, we do not mean to suggest that developed states ought, prudentially, to control migration. Perhaps unlimited immigration would be a good thing for developed states. On this issue, we take no stance.

• 2. In the next section, we identify one way in which, contrary to initial appearances, IA actually seems to constrain the options of would-be migrants. Although we demonstrate that the particular line of argument examined is ultimately unsatisfactory, it is worth noting why (and to what extent) the general strategy in question seems worth pursuing. Most obviously, if IA could be shown to constrain the options of would-be migrants, this would ‘unmask’ the practice. As noted in the Introduction, a large part of IA’s initial appeal is attributable to its apparently non-compulsory character and to the attendant contrast with the coercive and preventive forms of migration control with which we are more familiar. Indeed, IA’s proponents have defended the practice by claiming that it expands the opportunities of those with few valuable options (EUTF, 2022). If it could be shown that there is in fact a sense in which IA constrains the options of would-be migrants, this would undercut a significant aspect of its appeal.

Nevertheless, the significance of showing that IA constrains the option of immigrating is dependent on whether (and why) there is a right to immigrate. The grounds of the right to immigrate can be conceived in different ways. According to what we can call the modest conception of immigration rights, the right to immigrate is grounded in certain basic interests, such as the interests in security and subsistence (Blake, 2013: 125–129). On this view, one has a right to immigrate when immigrating is a necessary (or especially effective) means of satisfying such basic interests. This view is standardly taken to imply that the right to immigrate is possessed only by (some subset of) the poor and oppressed.

By contrast, according to what we can call the strong conception of immigration rights, the right to immigrate is grounded in certain non-basic interests, such as the interests in equality of opportunity and in extensive freedom of movement (Carens, 2013; Oberman, 2016). This view is standardly taken to imply that the right to immigrate is possessed not only by individuals who find themselves in especially dire circumstances, but rather by all human beings.

If one endorses the strong conception of immigration rights, evidence for the claim that IA constrains the option of immigrating constitutes evidence for the claim that IA constrains migrants’ rights. The latter claim would tell strongly against IA. On the other hand, if one endorses the modest conception, things look rather different. If IA is successful, it will address the basic needs of its recipients. It will thereby remove the grounds of any right to immigrate that those individuals might otherwise have possessed. If IA deprives its objects of a right to immigrate, then it can neither respect nor fail to respect that right. The right simply ceases to exist as a constraint that IA must observe.

In short, a strategy that seeks to criticize IA by demonstrating that it constrains the options of immigrants is dependent for its force on acceptance of the strong conception of migration rights. In this article, we do not attempt to adjudicate between this conception and the modest alternative. Rather, we examine (in the next section) one version of the critical strategy in question and reveal it to be ineffectual, regardless of the conception of immigration rights operating in the background.

• 3. We are not assuming that IA is actually effective at reducing migration. We acknowledge that evidence from the growing body of empirical literature that seeks to identify the effects of aid on migration is mixed. Among peer-reviewed studies, some suggest that development aid actually increases migration ⁵ (Berthélemy et al., 2009), while others maintain that it reduces migration (Lanati and Thiele, 2018). Some studies suggest that aid can effectively reduce migration if it is used specifically to improve the quality of governance and services (Gamso and Yuldashev, 2018), and some suggest that it can be effective in the long term (e.g. Dreher et al., 2019).

However, it is worth noting that even if it were established that IA is in fact ineffective at reducing migration, this would not render our article redundant. This is for two reasons. First, regardless of what the evidence shows, states are in fact providing IA, and they are doing so on a large scale. Given that this is so, IA demands a normative assessment. Second, if, in the face of what future evidence might establish, states misguidedly continued to provide IA in a futile attempt to control migration, one might argue that this should in fact be encouraged by third parties. Such encouragement might be appropriate if, although ineffective as a form of migration control, IA were effective as a form of development assistance. Under such circumstances, the argument would be that third parties should exploit donor states’ wrongheaded motivations in order to promote benign outcomes. ⁶ This argument has prima facie plausibility. By considering some of the morally relevant implications of providing IA, this article supplies resources that would be necessary for a conclusive assessment of the argument (though this is certainly not the article’s primary aim).

• 4. When evaluating IA as a form of development assistance, we advance an objection that is premised on the claim that developed states and their citizens have duties to assist the world’s poor. The success of our objection is conditional on the acceptability of this premise. Anyone who does not agree that there is a duty to assist the world’s poor will not endorse our objection. This obviously limits the reach of our objection, but we do not think it limits it unduly. The duty of assistance is discussed extensively in the literature on global justice, and while there is disagreement about its precise content and grounds, the duty’s existence is widely accepted. ⁷

• 5. In developing our evaluation, we focus on what we take to be the ‘core’ of IA. Accordingly, we set aside certain contingent or peripheral features that IA can sometimes – but need not – assume in practice and that have already been subjected to criticism. This means we set aside the claims that IA has sometimes been inadequately transparent; that it has not always reached its intended recipients; that is has sometimes been supplied to undemocratic regimes; and that donors have sometimes misidentified the root causes of migration (Raty & Shilhav, 2020).

In this context, it is also worth clarifying that we are not examining conditional offers of aid. Sometimes aid is provided on the condition that the recipient state accomplishes certain migration-related goals that have been specified by the donor (Raty & Shilhav, 2020). Although IA could, in practice, be offered conditionally, such conditionality is not a necessary feature of the practice and is not addressed here.

In short, our aim is to examine IA in its best possible light, shorn of any extraneous features or unsubstantiated limitations. If one wishes to establish whether a practice should be embraced or eschewed, one must not limit one’s attention to what the practice currently is, but must instead consider what it could be, if appropriately (re)designed.

Having set out these preliminary points, let us now turn to consider a first objection to IA.

Assessing Inducement Aid in its Capacity as a Form of Migration Control: The Message to Migrants Objection

As a first approximation of this objection, which we can call the message to migrants objection (MM), it might be said that IA sends a racist or xenophobic message to would-be migrants. ⁸ By seeking to induce foreigners to refrain from migrating, the members of donor states communicate the message that they want to limit the number of migrants living within their borders. Moreover, by devoting considerable resources to inducing foreigners to refrain from migrating, and thereby demonstrating that they are willing to make significant sacrifices in pursuit of this end, donor states reveal that they care a great deal about limiting the number of migrants living within their borders. According to the MM, would-be migrants may reasonably infer from this that they are held in low regard by policymakers in donor states. Moreover, if the donor state is democratic, then, since the policies of democratic states are supposed to reflect the wishes of the electorate, would-be migrants may reasonably infer that they are held in low regard by a majority – or at least a significant minority – of the donor state’s citizens.

The communication of such a message is bad in itself, simply in virtue of being insulting and disrespectful. It may also be bad in virtue of constraining important options of its recipients. One way of constraining someone’s options is to make the pursuit of those options more costly or difficult. A related way of constraining someone’s options is to make their pursuit appear more costly or difficult. If would-be migrants believe they are widely despised by citizens of the donor state, they may reasonably infer that any option they have of migrating will be difficult or costly to exercise. They may reasonably infer that integration will be a challenge and that everyday life will be made burdensome for them by natives. Relatedly, receipt of the message that one is despised may create a psychological obstacle to migration. As such, the message that migrants are held in low regard may be not only insulting but also option-constraining.

This is an initial formulation of the MM. Here is a possible reply. When a state provides IA, would-be migrants may presume that its provision reflects racist or xenophobic attitudes. But, at least in certain cases, this presumption will be unwarranted. This is because there are a variety of non-racist and non-xenophobic attitudes to which the donor state’s representatives may be responding. To illustrate, some citizens might wish to limit immigration not because they are racist, but because they believe immigration poses a threat to the economic wellbeing of disadvantaged domestic workers (Macedo, 2007). ⁹ Some other citizens may wish to limit immigration because they believe that doing so is necessary to preserve the distinctiveness of their national culture (Sides and Citrin, 2007). ¹⁰ To be sure, some citizens will undoubtedly have racist attitudes and will surely wish to limit immigration because of those attitudes. But, the reply continues, the provision of IA need not be taken as an expression of those attitudes, and it certainly need not suggest that such attitudes are commonplace among the donor state’s citizens. Therefore, the reply concludes, IA need not be insulting, and it need not communicate a message that, in virtue of a racist or xenophobic character, will make any option of migrating difficult or costly to exercise.

This reply may have force against our initial formulation of the MM. But it also helps to reveal that our initial formulation was unduly narrow. According to the objection’s initial formulation, IA might communicate attitudes that are racist or xenophobic. But if part of our concern with such attitudes reflects the fact that their communication is (1) insulting and disrespectful and/or (2) potentially option-constraining, we should widen the scope of our inquiry to include the communication of negative attitudes more generally. This is because racist and xenophobic attitudes are not the only kinds of negative attitudes the expression of which can be insulting or option-constraining. (As we move forward, it will be the ‘option-constraining’ variant of the MM that is in the spotlight, rather than the ‘disrespect’ variant, but it should be kept in mind that the reply we shall offer in fact speaks to both versions.)

Let us concede, then, that it may be wrongheaded to view IA as evidence of widespread racist or xenophobic attitudes among citizens of donor states. Let us grant that, in at least certain cases, it may be plausible to regard significant numbers of supporters of IA as motivated by a range of non-racist and non-xenophobic concerns. It is nevertheless the case that these supporters are motivated by a desire to limit immigration and by the notion that immigrants pose a threat to things that they value. As such, IA may plausibly communicate to would-be migrants the message that they are unwanted, and receipt of that message may constrain the options of would-be migrants in a manner that is similar (though perhaps less severe) to the manner in which receipt of a more overtly racist or xenophobic message may constrain their options. If would-be migrants believe that they are unwanted, and that they are unwanted because they are seen as threats to things that are highly valued by natives, they may reasonably infer that, should they migrate, natives will place obstacles on their path to integration, using legal and perhaps extra-legal methods to obstruct access to employment, accommodation, education, and/or other important goods. Moreover, the message that they are unwanted might add a psychological burden to migrating. If they believe that they are unwelcome, would-be migrants might find it more difficult to motivate themselves to undertake the difficult transition that migration involves.

This reformulated version of the MM encounters a powerful reply, which runs as follows. Although IA may communicate the message that migrants are unwanted, that message does not say anything new. After all, IA is not a state’s only or primary tool for controlling migration. A state’s primary tools for controlling migration are its admissions programme and the various policies it implements to enforce that programme. This fact is important because it suggests that whatever attitudes a community may have about migrants, those attitudes are already being expressed by the state’s other policies. We can capture this point by saying that, with regard to any message that it might communicate to would-be migrants, IA is expressively redundant.

Might IA result in the message being heard by audiences that would not otherwise have received it? If it could be shown that IA extends the reach of the message in this way, that would suffice to rebut the charge of expressive redundancy. It is, however, unlikely that IA extends the reach of the message. IA is targeted at areas with a history of migration to the donor state, and the residents of such areas are typically familiar with the donor state’s admissions policies. This is due to, for example, their own migration attempts or those of their social networks (Simon et al., 2018). This means that those who receive the message sent by IA are also already receiving comparable messages sent by other means, namely, via the donor state’s admissions programme.

The MM would also evade the charge of expressive redundancy if it could be shown that IA sends its message with greater force than alternative modes of communication. But this does not look like a promising escape route. Remember that the alternative modes of communication are coercive and preventive in character. It seems unlikely that aid says ‘stay out’ more forcefully than border enforcement or deportations. On the contrary, the message sent by IA is, if anything, weaker than the message sent by alternative forms of migration control. Whereas borders say simply ‘we will try to keep you out’, IA says, ‘we will try to keep you out, but in a way that addresses at least some of the reasons that ground your desire to enter’. Perhaps the latter message, taken in isolation, could still work to constrain the options of would-be migrants. But the latter message seems likely to produce a version of this effect that is weaker than that produced by the former, and when both messages are communicated simultaneously, the stronger effect of the former seems likely either to simply supplant the weaker effect of the latter or to be reduced by it.

These observations also enable us to pre-empt the suggestion that IA is non-redundant simply in virtue of repeating a message. Simple repetition may of course be expressively significant. If a racist shouts slurs at her neighbour every time she sees him, there is no reason to think that the tenth insult will sting less than the fifth. On the contrary, perhaps the insults become increasingly intolerable as they accumulate. But as we have now seen, IA does not merely repeat the message already communicated by other forms of migration control: the former sends a weaker message that may in fact dilute the message of the latter.

One final challenge to the expressive redundancy reply runs as follows. States’ traditional modes of migration control are more general than IA; they may distinguish between potential migrants on the basis of certain features, such as qualifications or family ties, but, typically, they do not draw more problematical distinctions, such as those based on nationality, ethnicity, or religion. By contrast, IA is country- or region-specific: its aim is to reduce migration from specific parts of the world. This means that the message IA communicates has a different content to the message communicated by states’ other policies. Whereas traditional modes of migration control send the message that, with certain exceptions, migrants in general are unwanted, IA sends the message that migrants from particular places are unwanted. Thus, IA is not expressively redundant.

We do not find this challenge persuasive. This is because it does not adequately characterize traditional modes of migration control. Although states’ admissions programmes may often appear to be quite general in character, in practice they are highly discriminatory, using features such as qualifications and class background as indirect measures to influence the national, ethnic, religious, and cultural composition of migration flows (Czaika and De Haas, 2013: 490). As one commentator notes, one upshot of these and other measures is that ‘the majority of Africans are de facto excluded from legal migration to Europe’ (Kleist, 2016: 5). In short, then, the alleged difference between IA and other forms of migration control is largely illusory. The former may be selective, but so too are the latter.

Ultimately, we believe that the MM is defeated by the expressive redundancy reply. Or rather, we believe that the MM is defeated by this reply when IA is used to supplement – rather than substitute – coercive and preventive modes of migration control. Since IA currently is used in a supplementary manner, it escapes the wrath of the MM. However, we said earlier that it is important to consider the practice of IA not only as it exists currently, but also as it could exist. Accordingly, we should acknowledge that IA could potentially be used as an alternative to other forms of migration control, rather than as a supplement. Used in this way, IA might look like an improvement over the status quo. It would represent a move away from the coercive and preventive modes of migration control that states have long relied upon. But an upshot of our analysis is that IA would then be vulnerable to the MM. This is because its proponents would no longer be able to point to alternative modes of migration control, acting as alternative forms of communication, that render its own expressive content redundant. If the argumentation of this section is sound, IA would then not be unambiguously option-enhancing – although it would surely constrain options less dramatically than coercive and preventive forms of migration control.

Assessing Inducement Aid in its Capacity as a Form of Development Assistance: The Message to Donors Objection

Let us now turn to a second objection to IA. Whereas the previous objection focused on the receipt of a particular message by would-be migrants, the objection to be considered here focuses on the receipt of a message by other developed states. According to this message, it is appropriate for donor states to use aid for self-interested purposes, even when this leaves the basic needs of the poorest unmet. The objection maintains that this message ascribes insufficient weight to the interests of the poorest and that it can have deleterious effects on the norms that govern the international domain. We refer to this objection as the message to donors objection (MD). In this section, we set out the objection in detail, explaining the various considerations from which it arises. For the sake of expositional ease, we shall start by expressing the objection quite boldly. Only once it has been set out fully will we describe how the objection must be qualified. We will then, in the subsequent section, identify various replies that may be given to the objection and explain how those replies might be answered. The upshot of our discussion is that the MD poses a genuine challenge to IA. In the conclusion, we briefly consider whether and how that challenge could be met.

Reducing migration is subjectively in the self-interest of many donor states and their citizens. In other words, many members of donor states believe that reducing migration is good for them. Therefore, when states use aid to reduce migration, they are using aid to further a goal that they perceive as self-interested. Third parties can correctly infer from this that donor states regard aid as (at least inter alia) an instrument for furthering their own ends, subjectively conceived. In this sense, IA communicates the message that it is appropriate for a donor state to use aid to further its own interests; the state providing aid signals to others its belief that such usage is acceptable. (For, why would the state in question use aid in this way if it did not regard such usage as acceptable?)

Of course, from the fact that reducing migration serves the (subjectively conceived) self-interest of donor states, it does not follow that IA does not also benefit the poor individuals who receive it. Often, it will be reasonable to expect that this aid will benefit those who receive it. Nor does it follow that donor states cannot be motivated, in part, by the prospect of benefitting these recipients. Donor states may provide IA in part to further their own interests and in part to benefit others. Moreover, where it exists, this dual motivation may be apparent to third parties. (In fact, third parties may attribute this dual motivation to donor states even in cases where it does not exist.) Perhaps, then, while IA communicates that aid may appropriately be used to pursue self-interested objectives, it need not communicate that this may be done when the objectives in question do not also align with the interests of the poor. A very charitable interpretation of IA’s message to donors is that (i) aid may appropriately be used to pursue self-interested objectives, (ii) provided that those objectives align with the interests of the poor. Since we ultimately want to suggest that a less benign message can plausibly be attributed to IA, we can adopt the very charitable formulation as our base interpretation. We will show that, even if this interpretation is plausible (in certain contexts) as a partial interpretation, it does not offer a complete account of what is communicated, and it has to be expanded. Once it is expanded to include extra details, the force of the MD becomes apparent.

If the proviso included in the charitable formulation is indeed part of the communicated message (at least in certain contexts), then IA may seem expressively innocuous. After all, there may be nothing objectionable about benefitting from the assistance that one provides to others, and, accordingly, there may be nothing objectionable about communicating that benefitting in this way is appropriate.

However, our description of IA’s message to donors is not yet complete. Notice that the charitable formulation refers to ‘the poor’ in general and does not draw any distinctions between different subgroups of this general category. More specifically, it does not distinguish between the moderately poor and the severely poor. However, the self-interested rationale for IA does draw this distinction. The self-interested rationale is that development assistance can reduce migration by improving the situation of those impoverished individuals who would otherwise be most likely to migrate. Now, the impoverished individuals who would otherwise be most likely to migrate are the moderately poor in particular, not the severely poor. To elaborate on this point, countries with low gross domestic product (GDP) per capita have lower rates of emigration than countries with higher GDP per capita (Hagen-Zanker, 2017; OECD, 2016), and, at a subnational level, international migrants tend not to come from the most deprived communities or the most deprived families (De Haas, 2007: 832). This is because migration entails high costs, such as the costs of travel and of foregone income, and these serve as a greater deterrent to the poorest. In addition, migrants must pay significant sums to obtain visas, work permits, or procure services in the recruitment process. Irregular migration entails additional costs, and these tend to be higher for low-skilled migrants – who often need to pay higher fees to labour recruiters for temporary and often informal jobs – than for high-skilled migrants (International Labour Organization, 2017; KNOMAD, 2017). Literature on ‘geographic poverty traps’ shows how migrants from more remote areas, who are also more likely to be poorer, also tend to face higher migration costs (Kraay and McKenzie, 2014).

In short, the self-interested rationale of reducing migration supports aiding the moderately poor, but not the severely poor. This has implications for the content of the message that IA can plausibly communicate to donors. If IA is directed primarily or exclusively at moderately poor countries and communities, then the base formulation of IA’s message to donors can be expanded as follows: (1) aid may appropriately be used to pursue self-interested objectives, (ii) provided that those objectives align with the interests of the poor, but (iii) regardless of whether the poor individuals whose interests are served are moderately or severely poor. Included within this message is the claim that aid may appropriately be used to pursue self-interested objectives (iii*) regardless of whether those objectives serve the interests of the severely poor. Going forward we shall focus on (iii*) rather than (iii).

This message can be interpreted in different ways. How the message is interpreted will depend, in part, on what meaning is attributed to the noun ‘aid’. This term may refer to any given aid initiative that a state operates, considered in isolation from other aid initiatives by which it may be supplemented, or it may refer to all of a state’s aid initiatives, considered holistically (its ‘aid programme’). Accordingly, the message may say that (i*) it is appropriate for a state to use one (of several) aid initiative(s) to pursue self-interested objectives . . ., or (i**) it is appropriate for a state to use its aid programme taken as a whole to pursue self-interested objectives . . .

The second version of the message is clearly morally problematical. As noted in the Preliminaries, we are assuming that rich states have a duty to assist the poor, and, under current circumstances at least, a state surely cannot discharge that duty if its aid programme, taken holistically, focuses primarily or exclusively on the moderately disadvantaged while largely or completely disregarding the interests of the poorest. ¹¹ Indeed, it is natural to suppose that the duty’s most stringent application is to the least advantaged. ¹² By contrast, it might be said that the first version of the message is unobjectionable, for it is acceptable for some aid initiatives to focus primarily or even exclusively on the moderately disadvantaged.

It may seem that IA communicates only the first version of these two messages and is therefore expressively unobjectionable. After all, it might be said, from any given aid initiative taken in isolation, one cannot infer a state’s attitudes or beliefs regarding what is and is not appropriate for an aid programme conceived holistically. Therefore, IA cannot reasonably be said to communicate anything about what is appropriate for an aid programme conceived holistically.

Notice, however, that simply substituting (i*) (rather than (i**)) for (i) is not sufficient to achieve an adequately complete interpretation of IA’s message. When interpreting the expressive content of any practice, it is important to consider the practice in its broader context. A general contextual feature of relevance here is that the basic needs of many severely poor individuals are currently not being met. While many wealthy states operate a wide range of aid initiatives, these are not sufficient (taken individually or collectively, or in conjunction with the policies of recipient states) to prevent widespread non-satisfaction of basic needs. In order to adequately account for this general background feature, we must further extend our formulation of IA’s message. According to the extended formulation, (i*) it is appropriate for a state to use an aid initiative to pursue self-interested objectives, (ii) provided that those objectives align with the interests of the poor, (iii*) but regardless of whether those objectives serve the interests of the severely poor, and (iv) even if the basic needs of the severely poor remain unmet.

This latest formulation of the message looks considerably more problematic than those that preceded it. Taken together, clauses (iii*) and (iv) sanction directing aid away from the poorest when the poorest have unmet basic needs. This demands a justification. If one is faced with a choice between alleviating the economic plight of the moderately poor or of the severely poor, the moral presumption must be that it is the severely poor who are to be aided. A satisfactory justification for instead directing aid to the moderately poor must point to the presence of some condition that can overturn this presumption. Such conditions are no doubt conceivable, yet none is included in the message communicated by IA. The only potential candidate justification included in that message is that directing aid away from the poorest is in the self-interest of the donor state. From the standpoint of morality, that is clearly inadequate.

We are now in a position to more fully understand the nature of the concern raised by the MD. The concern can be related to the kinds of norms and principles – the kind of international ethos – that IA might help to establish, maintain, or contest. If, as we have supposed, rich states and their citizens have a duty to assist the world’s poor – which would of course include the very poorest – the ethos by which the international community is governed should encourage the provision of such assistance. ¹³ The actions undertaken by members of this community should reflect the fact that the interests of the poor – including those of the poorest – have been accorded a suitably substantial weight. More specifically, rich states should be expected not to leave unmet the basic interests of the poorest simply in order to pursue alternative courses of action more conducive to the achievement of their own self-interest. ¹⁴ States should do what they can to uphold these expectations. ¹⁵

The MD maintains that, by providing IA, donor states act in a manner that is contrary to establishing and maintaining the relevant norms and principles (which, for ease of exposition, we can refer to as ‘cosmopolitan norms’). By signalling that it is appropriate to direct aid away from the poorest, thereby leaving the basic needs of the poorest unmet, and suggesting no justification other than that this serves their self-interested ends, donors communicate a particular conception of the weight that is to be accorded to the interests of the poorest, a conception that would surely be deemed inadequate by the relevant cosmopolitan standards. ¹⁶

The way in which such behaviour could be contrary to the establishment of relevant norms can be fleshed out by considering some insights from the constructivist school of International Relations, which examines the social processes by which norms emerge and become established within the international system (Florini, 1996: 367). The seminal article by Finnemore and Sikkink (1998) emphasizes that the establishment of norms relies on a critical mass of states acting as ‘norm leaders’. These states embrace and comply with the norms in question and socialize other states to do likewise. Thus, the establishment or ‘internalisation’ of cosmopolitan norms requires a preponderance of states embracing such norms. States that use aid for self-interested purposes, thereby leaving the basic needs of the poorest unmet, can be said to be neglecting their potential role as cosmopolitan norm leaders or norm followers. In doing so, they may limit the extent to which the relevant norms are established or maintained. A great deal of research examines how revisionists or challengers, or simply states that choose not to comply with international norms, may contribute to the regression or decay of those norms (see Deitelhoff and Zimmermann, 2020, for a review).

These considerations return us to a recurring theme of our article. If IA is contrary to the establishment and maintenance of cosmopolitan norms, it should not be regarded as unambiguously option-enhancing. This is because (1) the quantity and quality of the poor’s options are inversely correlated with the level of their poverty, (2) effective assistance reduces poverty and (3) robust cosmopolitan norms support the provision of effective assistance to the poor. (This is part of what it means to say that the norms in question are robustly cosmopolitan.) If IA threatens cosmopolitan norms, it thereby threatens the kind of assistance that can be expected to enhance the options of the poorest.

We said that we would start by setting out the MD in quite bold terms and that it would ultimately have to be qualified. As we have formulated it so far, the MD says that IA communicates a message with a very specific content. Is the claim that each and every donor must inevitably receive this message and that its content will be interpreted in exactly the same way by all recipients? This hardly seems credible. What the foregoing argument does a better job of establishing is that the message can reasonably be interpreted in the way that it has been described. If any recipient understands the message in that way, a proponent of IA cannot plausibly claim, in defence of the practice, that the recipient has misinterpreted the message, that the fault lies with the ‘listener’ and not with the ‘speaker’. Of course, it is then a further question whether states will actually interpret the message in the specified manner and a further question still whether the message will be interpreted in that way by a number of states sufficient for the predicted effect on international norms to materialize. Qualified in this way, the MD claims that, by acting in a manner that can reasonably be interpreted as insufficiently attentive to the interests of the poorest, donor states are irresponsibly neglecting their role as potential cosmopolitan norm leaders or followers.

The MD: Possible Replies

Having now delineated and qualified the MD, let us proceed to consider three possible replies. The first is as follows: although the moderately poor are the intended beneficiaries of IA, the severely poor will also gain. The severely poor may benefit directly (e.g. because there are pockets of severely poor individuals in moderately poor countries, and these individuals, along with moderately poor individuals, receive the benefits that are aimed at the latter) or indirectly (e.g. because the severely poor gain when their moderately poor compatriots refrain from emigrating, set up businesses, and so forth).

In response, given that the MD is expressive in character, the issue of whom aid is expressly aimed at is more relevant to its assessment than the issue of who actually ends up benefitting. It is problematical for donors to communicate the message that they are expressly targeting aid at the moderately poor, for self-interested purposes, while the basic needs of the very poorest remain unmet, even if some of the poorest end up benefitting incidentally. Of course, if benefitting the moderately poor were the optimal way of benefitting the severely poor, this message would not be communicated. But this hardly seems likely. Consider two points. First, although it is true that the moderately poor countries and regions at which IA is targeted have pockets of severely poor individuals, we should not lose sight of the fact, already emphasized, that the self-interested rationale for IA gives donors little reason to target the very poorest countries and regions. Second, given that IA is not targeted at highly skilled workers (whom rich countries are typically eager to recruit), it cannot be expected to adequately address the phenomenon known as the ‘brain drain’, whereby those with advanced educations emigrate en masse, leaving their less advantaged compatriots behind.

A second reply, which we can call the realist reply (RR), runs as follows. While IA may communicate a message that ascribes insufficient weight to the interests of the poorest, it is also a potentially potent advertisement for how aid can serve the self-interested ends of donors. Since self-interest is a powerful motivator, receptive states may respond by increasing overall aid expenditure. To be sure, the self-interested rationale behind IA gives donors a reason to direct much of their aid away from the severely poor, but losses to this group could potentially be adequately compensated by concomitant gains to the moderately poor.

The RR can acknowledge that neglecting the poorest for self-interested reasons, and publicly endorsing such neglect, is contrary to the duty of assistance – for it can acknowledge that duty demands assistance be offered to both the severely and moderately poor – but it can insist that the likely outcomes of any policy must be compared not to a fully just state of affairs (which will no doubt elude us for the foreseeable future), but rather to the likely outcomes that would otherwise obtain in the highly imperfect world in which we currently live. The reply contends that the latter, more appropriate, comparison may in fact favour IA.

We will now offer five rejoinders to the RR. Let us note in advance that none of these is sufficient in itself to defeat the RR. (In that sense, they are each only partial rejoinders.) But, taken in conjunction, they expose the limits of that reply. First, it is widely believed (e.g. by both prioritarians and egalitarians) that gains and losses to the least advantaged must be weighted more heavily than gains and losses to those who are comparatively better-off (O’Neill, 2008; Parfit, 1997). This means that a gain to the moderately poor is not a full substitute for a comparable gain lost by the severely poor. ¹⁷ Suppose that any gain/loss to the moderately poor is worth 0.75 of any gain/loss to the severely poor. Then, to compensate for 10 losses suffered by the severely poor, the moderately poor would have to secure in excess of 13 gains. In short, for the RR to succeed on its own terms, IA would have to give more to the moderately poor than it takes from the severely poor.

Second, IA is not vindicated by the mere possibility that it could produce better outcomes than those that would otherwise obtain in our unjust world. If the MD succeeds in showing that IA can reasonably be interpreted as ascribing insufficient weight to the interests of the poorest, and that it thereby risks publicly sanctioning behaviour contrary to what the duty of assistance demands, that places a significant burden of proof on those who would defend IA. Those who wish to defend it cannot simply assert that IA would produce better outcomes or content themselves with the thought that IA might produce better outcomes. If the MD successfully establishes that IA can reasonably be interpreted as communicating a message consisting of clauses (i*), (ii), (iii*) and (iv), this throws down a challenge that demands a more serious response.

Third, it bears emphasizing that the MD is about fairly general norms. It is concerned with the establishment and maintenance of an international ethos that ascribes substantial weight to the interests of the poorest and that upholds important cosmopolitan standards to which states are expected to conform. An ethos of this kind would be broad in scope. It would influence states not only in their formulation of an aid programme, but also in their design of policies in various other areas, such as trade and climate. In a world governed by a robust cosmopolitan ethos, policies in these areas could be expected to be considerably more favourable to the poorest than they would be in a world without such an ethos. This means that the gains IA brings to the moderately poor must be compared not only to losses that the severely poor suffer in terms of forgone aid, but also to the losses that they might suffer in other areas as a result of a degraded international ethos.

Fourth, even if IA could produce gains for the moderately poor that were large enough to represent an all-things-considered improvement on the alternative situation that would otherwise obtain, this would not establish that IA is a purely option-enhancing enterprise. On the contrary, it would concede that IA enhances the options of the moderately poor at the expense of the options of the severely poor.

Our final rejoinder merits more attention than we have space to offer. Perhaps we should refer to it not merely as a partial rejoinder but as a potential partial rejoinder that could perhaps supplement the others if fully developed. This potential partial rejoinder is as follows. While the RR enjoins us to take seriously what it is possible to achieve in our radically non-ideal world populated by highly imperfectly motivated actors, we must also take care not to close off promising routes of escape from this condition. The status quo is clearly grossly unsatisfactory from the standpoint of morality. There are many severely poor people in the world, and, as we have seen, the self-interested rationale behind IA does not give donors a reason to help them. An alternative, as suggested above, is to strive to establish other-regarding cosmopolitan norms. These norms expect states to sacrifice some of their own ends when doing so is necessary to secure the basic needs of the poorest. ¹⁸ A potential problem with the RR is that the self-interested approach that it seeks to vindicate may be in tension with the establishment of such norms. This would be an instance of the more general problem that extrinsic incentives can sometimes ‘crowd out’ intrinsic motivation. ¹⁹ In this case, the concern is that if members of affluent states are encouraged to assess international policies in terms of what those policies can do for them, altruistic deliberation might be displaced by an altogether more self-interested perspective, which would be contrary to the establishment of cosmopolitan norms, and thus to the interests of the poorest.

Of course, it would be hopelessly quixotic to suggest that no cosmopolitan progress can be made until self-interested motives are completely purged from the international scene. But we can acknowledge this point while also thinking that the particularities of IA make its public endorsement of pursuing self-interest at the expense of the poorest especially concerning. We shall highlight some of these particularities in the course of responding to a final potential reply to the MD.

This final reply invokes concerns about expressive redundancy. This reply, like the analogous one that we pressed against the MM, claims that any message communicated by IA is unlikely to say anything new. IA is not the only form of aid that appears to be used to further self-interested objectives while leaving the basic needs of the poorest unmet. Thus, it might be said that any message IA sends about the appropriate weight to be accorded to the interests of the poorest has already been communicated by other forms of aid.

Although we argued that the message which IA communicates to would-be migrants is likely to be redundant, several considerations challenge the suggestion that the message communicated to donors is also redundant. Notice, first, that with regard to many forms of aid and many instances of aid provision, the donor’s motive will not be entirely perspicuous. One subset of recipients might reasonably interpret the motive in one way (e.g. as self-interested); another subset might reasonably interpret it in a different way (e.g. as other-regarding); and another subset might see some degree of plausibility in several different possible motives. ²⁰ In this context, note that the evidence adduced to support the claim that particular instances of aid provision must be motivated by self-interest is not always decisive. For example, some states send a disproportionate amount of aid to their former colonies, and this is sometimes said to reveal that donors are motivated by political and strategic considerations (Alesina and Dollar, 2000). But donors’ preferences for former colonies can be explained in other ways. For instance, donors might have more information about former colonies than about other developing states, and this may reduce the costs associated with the monitoring of aid provision (Qian, 2015: 298). In short, other potentially self-interested forms of aid may be insufficiently clear-cut in their rationale to render IA expressively redundant in relation to all relevant audiences.

Relatedly, when taken together, multiple forms of aid may ‘speak’ with greater clarity than when each is considered in isolation. For a given audience, IA might constitute an additional piece of evidence for the belief that some other forms of aid are motivated by donor self-interest (and vice versa). In other words, rather than rendering each other expressively redundant, different forms of aid might contribute to clarifying the message that each communicates. If this is true, the case against IA is not weakened by the existence of other potentially donor-serving aid practices, but rather strengthened.

Finally, there is some evidence to suggest that IA is becoming progressively more dominant. To the extent that this is so, it can be expected to ‘speak’ with a relatively high degree of authority. Let us finish by elaborating on this claim that inducement is becoming more dominant. Increasingly, extant development assistance is being relabelled as IA, and formerly ad hoc activities linking migration and development assistance are becoming part of official strategies with the explicit aim of reducing migration (Knoll and Sheriff, 2017: 25). This is a result of broader political developments. In Europe, the recent refugee emergency ‘has pushed substantive engagement on migration up the priority ladder for development cooperation’, and experts, government officials, and members of civil society all note that ‘political expectations for [Official Development Assistance] spending to reduce migration flows are high’ (Knoll and Sheriff, 2017: 24). This policy shift is in line with changing political rhetoric. Contrary to expectations, the resurgence of populism has not necessarily led to an anti-aid stance, but to a rebranding of aid as a tool to limit the admission of immigrants and refugees (Bayram and Thomson, 2021). If IA is becoming a particularly popular and dominant form of aid (Kipp, 2018), then any message communicated by IA can be expected to command greater authority than those communicated by relatively marginal forms of aid.

Conclusion

Let us finish by drawing together some of the main strands of our discussion and by positing some conclusions. In its capacity as a form of migration control, IA has prima facie appeal. From a normative perspective, it certainly looks more attractive than the preventive and coercive forms of migration control that it exists alongside. Moreover, so long as it continues to merely supplement these more traditional forms of migration control, IA, we have argued, will be immune to the MM. This is because preventive and coercive forms of migration control render IA’s message to migrants expressively redundant. Of course, one can imagine affluent states one day dispensing with preventive and coercive forms of migration control and operating migration regimes that rely exclusively on IA. IA would then become vulnerable to the MM, but migration regimes of this kind would surely be superior, all-things-considered, to those currently in operation.

A complete assessment must also consider how IA fares as a form of development assistance. In this capacity, we have acknowledged that IA might have considerable benefits. This is because IA identifies and advertises a powerful self-interested rationale for states to provide aid. However, while this self-interested rationale supports providing aid to the moderately poor, it does not support aiding the poorest. Moreover, by publicly sanctioning activities that neglect the least advantaged for self-interested purposes, IA sends a message that is contrary to the establishment and maintenance of important international norms that have a potentially valuable role to play in facilitating cosmopolitan progress.

If these concerns are well-founded, then, initial appearances notwithstanding, IA is not unambiguously option-enhancing. It may preserve the options of migrants (qua migrants) and enhance the options of the moderately poor, but by threatening important international norms, it may also put in jeopardy some of the most valuable options of the very poorest.

Does our analysis show that IA must necessarily be abandoned? It does not. The MD claims that, by communicating what can reasonably be interpreted as an anti-cosmopolitan message, donors irresponsibly neglect their role as potential norm leaders or norm followers. However, we have noted that how an aid initiative is interpreted may be influenced by other policies – including other aid initiatives – that the state in question undertakes. Various policies can reinforce each other’s messages, but they can also potentially counteract each other’s messages. Aid provided alongside IA could, in principle, communicate a very different message or even prompt relevant audiences to interpret IA differently.

Aid can be given for other-regarding reasons, and these reasons can be expressed with greater or lesser clarity. Donors can make efforts to signal their altruistic motivations. They can do this, for example, by ensuring that aid is directed to the poorest regions and that it clearly responds to recipient needs. ²¹ If IA were supplemented by manifestly other-regarding donations, its potentially pernicious message could perhaps be offset – or even reframed. If this could be done, there may be no compelling reason to terminate IA.

But we do not pretend to have settled this question here. We offer our assessment as the first (normative) word on the topic, not the last. Our hope is that the argument we have pressed will encourage other normative theorists to engage with these matters, thereby supplementing the valuable work already being done in the social sciences.