OPEN BORDERS AND THE IDEALITY OF APPROACHES: AN ANALYSIS OF JOSEPH CARENS’ CRITIQUE OF THE CONVENTIONAL VIEW REGARDING IMMIGRATION

Do liberal states have a moral duty to admit immigrants? According to what has been called the “conventional view”, this question is to be answered in the negative. One of the most prominent critics of the conventional view is Joseph Carens. In the past 30 years Carens’ contributions to the open borders debate have gradually taken on a different complexion. This is explained by the varying “ideality” of his approaches. Sometimes Carens attempts to figure out what states would be obliged to do under otherwise perfectly just conditions (i.e., he attempts to establish an ideal). At other times, he is more interested in what to do, given the (not fully just) world that we actually live in. In my view, the relevance of the ideal/non-ideal theory debate to the open borders debate (and the ethics of migration more generally) has not yet received sufficient attention. My aim in this paper therefore is to show in detail how Carens’ varying approaches affect his critique of the conventional view. To this effect I analyse three of his papers: “Aliens and Citizens: The Case for Open Borders” (1987), “Realistic and Idealistic Approaches to the Ethics of Migration” (1996), and “Who Should get in? The Ethics of Immigration Admissions” (2003).


Introduction
Do liberal states have a moral duty to admit immigrants? According to what has been called the "conventional view", this question is to be answered in the negative. Liberal states do not have a strong duty to admit immigrants; or at least they do not have a duty to admit all immigrants. In this respect, liberal states have been claimed to resemble clubs (Walzer 1984, 14-39; see also Wellman 2008). Just as the members of a club may award or reject membership based on (almost) any criteria, we should consider liberal states to be free to award or deny membership based on (almost) any criteria as well. That is, how many immigrants they admit and which ones they admit, is (largely) up to them. Michael Walzer, one of the conventional view's main proponents, puts it as follows: The distribution of membership is not pervasively subject to the constraints of justice. Across a considerable range of the decisions that are made, states are simply free to take in strangers (or not)much as they are free, leaving aside the claims of the needy, to share their wealth with foreign friends, to honor the achievements of foreign artists, scholars, and scientists, to choose their trading partners, and to enter into collective security arrangements with foreign states. (Walzer 1984, 61) The conventional view arguably reflects how most people in Western societies think about immigration. Nevertheless, in recent years it has provoked strong criticism. Scholars have argued that liberal states do have a strong moral duty to admit immigrants; e.g., because this is implied by the democratic theory of popular sovereignty (e.g., Abizadeh 2008) or because more open borders are a way of compensating for injustices such as poverty or human rights violations (e.g., Wilcox 2007).
One of the most prominent critics of the conventional view is Joseph Carens. Carens believes that borders should be (far) more open than they currently are. His argument for this claim rests on the idea that citizenship, though of tremendous influence on people's prospects of having a fulfilled life, is not the result of personal efforts or achievements. It is rather something that we are "born into". If one is lucky, one comes into the world within the borders of a state such as Austria. But one may as well be born in Bangladesh, or Sudan, or North Korea, and be doomed to a life in poverty. This does not seem fair. Thus, for Carens citizenship is "the modern equivalent of feudal privilege " (1987; see also, e.g., 1996, 169). One prima facie puzzling feature of Carens' critique of the conventional view is that it comes in different degrees, or is situated on different levels. In some of his publications Carens rejects the conventional view altogether, advocating radical openings of our borders. On other occasions, in contrast, he suggests only minor departures from the status quo, e.g., rethinking some of the selection criteria for immigrants. This difference is explained by the fact that Carens considers the open borders debate from different perspectives. Sometimes he tries to figure out what states would be obliged to do under otherwise perfectly just conditions. He tries to establish an ideal; something that we can orient ourselves by. At other times Carens is more interested in what to do, given the world we live ina world which involves injustices, in which many policies must be regarded as infeasible, in which there are limits as to what we can demand of people, etc. Put differently, his approach varies in how "ideal" or "nonideal" it is (see Valentini 2012).
In my view, the relevance of the ideal/non-ideal theory debate to the open borders debate (and the ethics of migration more generally) has not yet received sufficient attention. My aim in this paper therefore is to show in detail how Carens' critique of the conventional view puts on a different complexion depending on the "ideality" of the approach that he takes. To this effect I will analyse three of Carens' papers. In the second chapter I will investigate what is probably his most famous article, "Aliens and Citizens: The Case for Open Borders", dating back to 1987. In the third chapter I will analyze Carens' 1996 article "Realistic and Idealistic Approaches to the Ethics of Migration". Finally, in the fourth chapter, I will be concerned with a more recent article entitled "Who Should get in? The Ethics of Immigration Admissions" (2003). Before going in medias res, however, some words are in order regarding the distinction between ideal and non-ideal theory.

Ideal vs. Non-Ideal Theory
The distinction between ideal and non-ideal theorizing has been drawn in a variety of different ways (e.g. Rawls 1971, 8-9, 244-248;Farelly 2007, 844; for an overview see Valentini 2012). In "Realistic and Idealistic Approaches to the Ethics of Migration" (1996, 157) Carens himself provides a definition. In his understanding, ideal and non-ideal theorizing differ with respect to how large they allow the "gap" between is and ought to be.
According to non-ideal theory (e.g., Farelly 2007;Galston 2010;Gaus 2017;Horton 2010), the gap between is and ought must not be too big. This is because morality is taken to be essentially action-guiding, and if it were to prescribe actions which cannot be performed here and now (because they demand too much in terms of human psychology, because they are politically infeasible, etc.) then it could not fulfill this function. It would miss its point. Ideal theorists (e.g., Cohen 2009;Estlund 2017) reject this claim. They believe that what ought to be is widely independent of what is, often because of what has been called the problem of "adaptive preference formation". Suppose one allowed empirical facts to restrict moral prescriptions in the way non-ideal theorists do. Then, the worry goes, we would often be satisfied with too little (in terms of justice); we would accept the status quo even if it were somewhat or considerably unjust.
Carens illustrates the problem with extremely non-ideal approaches by the example of slavery in America of the 17 th and 18 th century (1996,(164)(165). As slavery was a stable social institution by this time, as there was only little emotional identification with the slaves, as slave-owners had a strong interest in maintaining things as they were, and as abolishing slavery was politically infeasible, non-ideal theorists would be forced to say that slavery was morally permissible. But this conclusion is unacceptable. Thus, ideal theorists argue, philosophers need not (or only somewhat) be concerned with actual empirical facts. Moral considerations should be guided by ideals; they should be about what is possible under ideal (rather than actual) circumstances.
In the following quote Farelly summarizes the main problems of each of the two approaches, both the ideal and the non-ideal one: At the extreme of fact-insensitivity (what we can call extreme ideal theory), one runs the risk of invoking an account of justice that fails to function as an adequate guide for our collective action in the real, non-ideal world. At the opposite end of the spectrum is the danger that all existing constraints (even those imposed by an unjust social structure) are taken as legitimate constraints and thus justice simply reaffirms the status quo (Farelly 2007, 846). By speaking of "extremes" and of the "ends of a spectrum", Farelly's quote points to an important qualification. Only few philosophers (but see, e.g., Cohen 2009 for the ideal side) advocate ideal or non-ideal theorizing in their extreme forms. Commonly, what is claimed is not that empirical facts restrict moral prescriptions in the sense of strictly determining them or that empirical facts do not have any implications for the validity of moral prescriptions at all, but rather that the ought depends on the is to a certain (smaller or larger) degree. Thus, on most of the accounts that have been proposed, and certainly on the account that is assumed here, the distinction between ideal and non-ideal theory is best thought of as a spectrum (Hamlin and Stemplowska 2012;Mason 2004;Valentini 2012).
Discussions about ideal and non-ideal theory have often focused on which level of analysis is the most appropriate one (see, e.g., Cohen 2009;Farelly 2007). When we reason about justice or morality in general should we take a more ideal or a more non-ideal approach? Some philosophers have suggested, however, that each of these levels of analysis is legitimate. They do not exclude but rather complement each other. For some purposes more ideal and for other purposes more non-ideal approaches are appropriate (e.g. Hamlin and Stemplowska 2012;Mason 2004;Rawls 1971;Valentini 2012). This is the view Carens subscribes toand that I find most plausibleas well.
In "Realistic and idealistic approaches to the ethics of migration" (1996,(168)(169) Carens explicitly claims that regarding questions of migration there is no such thing as a correct degree of idealization. If one is interested in establishing action-guiding prescriptions one should look at things from a non-ideal perspective. If one's aim is to evaluate certain institutions or practices or to establish long-term goals, in contrast, an ideal approach is more appropriate: […] what is at stake here is more a matter of differing sensibilities and strategies of inquiry than of logically incompatible positions.
[…] Each approach has something important to contribute to the ethics of migration. (Carens 1996, 156-157) Ultimately what is needed is a full range of reflections, each selfconscious and explicit about its own purposes and presuppositions. There is no uniquely satisfying perspective on the ethics of migration. (Carens 1996, 169) This is a very important point for understanding Carens' work. As mentioned, he has looked at the open borders debate both from ideal and non-ideal perspectives. One may be led to think that this is explained by revision, i.e., by the fact that Carens first considered ideal approaches to be more appropriate, but later came to the conclusion that it is better to take a non-ideal stance. However, in light of the above remarks this might not be true. Carens seems to believe that each of the different levels of analysis has its merits. Each of them helps us to gain a better understanding of what states ought to do with regard to admitting immigrants.

Aliens and Citizens: The Case for Open Borders
"Aliens and Citizens: The Case for Open Borders" marks the beginning of Carens' engagement with the ethics of migration. Carens' perspective in this article is mostly the perspective of extreme ideal theory. What he is interested in is not what states are morally obliged to do, given the circumstances that actually obtain (although such considerations play some role in his discussion of Rawls). The paper is rather about justice in an ideal sense.
From a largely ideal point of view the conventional view turns out to be wrong, according to Carens. Liberal states do have a moral duty to admit immigrants, and they have this duty not only with respect to some of their would-be citizens, but with respect to nearly all of them. Put differently,

Carens believes that borders should be widely open: "[…] borders should generally be open and […]
people should normally be free to leave their country of origin and settle in another, subject only to the sorts of constraints that bind current citizens in their new country" (1987,251). 1 Carens considers what he takes to be the three main approaches to political theory: property rights theories, John Rawls' theory of justice, and utilitarianism. He attempts to show that each of these theories implies his above claim, i.e., that borders should be much more open than they currently are.

Property Rights Theories
In arguing for the conventional view people often stress the fact that the state they are citizens of is their state. This might be interpreted as an appeal to property rights, or more specifically, to collective property rights. A state, the argument goes, is the citizens' collective property. In some sense it is owned by them. Therefore, the citizens can exclude whomever they want.
Is this line of reasoning convincing? Does an appeal to property rights really lend support to the conventional view? Carens denies that this is actually the case. In particular, he attempts to refute the above argument by the example of the most prominent contemporary proponent of property rights theory, namely Robert Nozick (1974).
Nozick starts from the assumption that people have certain natural rights, including the rights to acquire and use property and to enter into voluntary exchanges. Moreover, he assumes that the sole purpose of the state is to protect people on a certain territory from violations of these rights. On such a view, Carens argues, states do not have a right to restrict immigration (1987,(253)(254). Suppose a US farmer hires a Mexican worker. Since this is a voluntary exchange, the state must not prevent the farmer and the worker from doing so; it is even obliged to protect their right against other people's interferences. But suppose the Mexican worker does not have any job offer. Is Nozick's minimal state at least justified in excluding him under these circumstances? No, Carens argues. As long as the Mexican does not violate the natural rights of other individuals the state must not exclude him in this case either. On Nozick's account, who enters a state is none of the state's business. It exclusively depends on the individuals living within the state's borders. They can admit and exclude whomever they want, citizen or non-citizen. The state, however, is not justified in restricting immigration.

Rawls' Theory of Justice
In "A Theory of Justice" (1971) Rawls explicitly distinguishes between ideal and non-ideal approaches to justice. Although Rawls himself does not discuss questions of immigration (starting from his assumptions they simply do not arise), Carens believes that his theory can be expanded to cover such issues.
Rawls attempts to justify his principles of justice by reference to a hypothetical and ahistoric formation of a contract. The terms of this contract are negotiated behind a "veil of ignorance", i.e., the parties of the contract do not have knowledge about their class, their sex, their race, their goals and so on. Carens (1987, 256) argues that one's citizenship should also be among the things that are covered by the "veil of ignorance". After all, citizenship is exactly one of those contingent features Rawls attempted to get rid of to promote impartiality. According to Carens, the "veiling" of one's citizenship would not affect the general terms of the contract. The parties would still agree on the principles set out by Rawls: the first principle that guarantees an equal set of basic liberties for all people; and the second principle, according to which social and economic inequalities are justified only if they are to the greatest benefit of the worst-off and the positions attached to them are open to all people under fair conditions. That said, the basic liberties of the first principle would now also involve the principle of freedom of movement between states. This is because even under ideal conditions people can have reason to migrate from one country to another. They can fall in love with a citizen of a foreign country; they may want to migrate for economic or religious reasons; and so on. Since the parties of the hypothetical contract ex hypothesi reason from the perspective of those that are made worst-off by a restriction they would therefore agree that the right to migrate should be one of the basic liberties (Carens 1987, 259-262).
The above holds true, according to Carens, even if we take into consideration a qualification made by Rawls. Rawls maintains that a liberty may be restricted if it threatens public order and thus other liberties. At first sight it might seem as if an unrestricted right to migrate inevitably threatens public order. However, Carens argues that this is actually very unlikely. Under ideal circumstances citizens would not protest against just regulations and states would be just as well (which means that the likelihood of mass migration would be low).
In addition, Carens (1987) points out that Rawls himself recommended great caution in applying the above "public order restriction". Starting from his non-ideal theory (where historical contingencies and actual injustices are taken into account) at least some restrictions on immigration turn out to be justified. Carens (1987) considers three cases: first, the case of people who aim at overthrowing just institutions; second, the case of mass immigration into one state; and third, the case in which the right of freedom of movement is restricted for the sake of economic gains.
According to Carens, in the first and in the second case some restrictions on immigration are justifiable. Both people threatening national security and mass immigration have the potential to lead to a breakdown of public order. In the third case, restrictions may be justifiable in principle. In ideal theory Rawls grants basic liberties priority over social and economic gains. This "lexical priority" is weakened in non-ideal theory, however. If the economic gains at issue benefit the worst-off and promote justice, Rawls allows them to override basic liberties for some time. Carens (1987, 262-263) argues, however, that these conditions are probably not sufficiently met in the case of restricting immigration. First, it is not clear how restricting immigration should benefit the worst-off. It seems, to the contrary, that immigration itself in many cases benefits the worst-off economically. Second, even if the first point did not hold, we would very likely have more effective means of improving the situation of the worstoff than by restricting immigration.
To sum up Carens' interpretation, Rawls' ideal theory implies that states are not justified in restricting immigration and that, consequently, open borders should be our long-term goal. Non-ideal theory implies that restrictions are justifiable, but only in a small number of cases.

Utilitarianism
According to utilitarianism, the moral rightness of an action depends on its contributing to the maximization of utility (defined, e.g., as pleasure, preferences or interests). At first sight, restrictions on immigration may seem justified on utilitarian grounds. After all, do not at least some citizens suffer economically from immigrants entering their state?
In response to this argument, Carens points out that according to utilitarianism, "everyone is to count for one and no one for more than one" (1987,263). This means that not only the utility of some of the state's current citizens has to be taken into consideration, but the utility of all of its citizens and, even more importantly, all aliens as well. Since there are probably citizens who benefit from more open borders, since there are surely very many aliens who do so, and since the free mobility of labour is said to be to everyone's economic advantage, restrictions do not seem justified from a utilitarian point of view either. Restrictions may only be justified, Carens argues, if one takes into consideration certain unreasonable, shortsighted or evil pleasures or displeasures (like the fear of one's culture becoming suppressed or racial prejudice). Even under this (supposedly wrongheaded) assumption, however, the restrictions would only be minor compared to those that are actually in place (Carens 1987, 263-264).
Since significant restrictions on immigration can be justified on neither property rights theories, nor Rawlsian, nor utilitarian grounds, Carens concludes that it is very unlikely that the conventional view is true. States do have a moral duty to admit immigrants. Borders should be far more open than they currently are. This is the ideal we ought to strive for (Carens 1987, 270).

Realistic and Idealistic Approaches to the Ethics of Migration
"Realistic and Idealistic Approaches to the Ethics of Migration" focuses explicitly on the distinction between ideal and non-ideal theorizing. In this way, the article differs from most of Carens' other publications. Instead of arguing for a particular conclusion regarding the open borders debate, Carens is rather interested in what conclusions might be implied by taking more ideal or more non-ideal approaches, and how one should conceive of the relation between ideal and non-ideal theorizing.

Empirical Restrictions
According to non-ideal theory, moral prescriptions are restricted by empirical facts. This raises the question of what kinds of empirical facts are relevant. With respect to the ethics of migration, Carens (1996, 158) argues that the most important kinds of facts are (1) institutional, (2) behavioural and (3) political facts.
The most significant institutional restriction is claimed to arise from the existence of a system of sovereign and independent states. According to Carens, this system is so firmly established that if moral prescriptions require substantial departures from it then they cannot function to guide our actions: "An ethics of migration that requires abolition or even radical transformation of the state system is not a morality that can help us to determine what is to be done in practice" (Carens 1996, 159). One implication of this acknowledgment of the modern state system is that one also has to acknowledge states' sovereignty with regard to immigration. One has to acknowledge that states have the authority to exclude aliens, or at least most aliens, as they like (Carens 1996, 159-160).
The second kind of facts that should be taken into consideration within non-ideal approaches to the ethics of migration are facts about human behaviour. According to Carens, moral prescriptions must not be too demanding. They must not prescribe what most people most of the time are unable to do, or cannot be realistically expected to do. With respect to the open borders debate this means, e.g., that states cannot be judged by standards such as admitting all refugees that seek asylum. Such a standard would simply be too ambitious. It would place too big of a burden on the state and its citizens (Carens 1996, 158-159).

The third kind of restriction Carens discusses in "Realistic and Idealistic
Approaches to the Ethics of Migration" are political restrictions. From a non-ideal point of view morality should confine itself to politically feasible prescriptions. Opening all gates between all states may be a noble ideal, but policies such as this do not have a chance of being implemented. On non-ideal accounts elaborating them and discussing them appears to be a waste of time (Carens 1996, 159-160). Furthermore, one has to keep in mind the risk of a backlash against immigrants and refugees by the current citizens of a state. If restrictions on immigration are weakened, citizens can easily get the impression that borders are "out of control" and demand regulations that are even harsher than the prior ones in force.
One might argue that such a reaction is unjust, or racist, or unreasonable. But that's beside the point, according to Carens (1996, 160): "There is no point in wasting time considering whether the popular reaction is racist or whether the politicians might be able to prevent such a reaction if they expended vast amounts of political capital to do so." This stands in stark contrast to Carens' view in "Aliens and Citizens", as discussed in the previous section. Looking at the issue from an idealistic perspective, Carens there argued that "evil pleasures" or "evil reactions" should not be taken into consideration in determining what we morally ought to do (see his discussion of utilitarianism).

Considerations about Effectiveness
In addition to the above three restrictions, Carens argues that the strength of our moral obligations is also affected by their effectiveness. Having effectsbeing action-guidingis what morality is all about, according to the non-ideal approach. So the less effective a moral prescription is, the weaker it should be considered to be. What makes moral prescriptions effective? Carens (1998, 160) discusses three kinds of preconditions.
The first precondition is psychological. In order for morality to be effective, there has to be some kind of emotional identification with the moral subject. The stronger the emotional identification is, the more effective and thus the stronger our moral obligations are. Since people typically identify more strongly with citizens of their own country than with aliens, this means that we have stronger moral obligations towards the former than the latter. The state is morally justified to weigh the interests of its citizens higher than the interests of those who want to become part of the state (Carens 1996, 160-161).
The second precondition for an effective morality is sociological. Carens points out that moral prescriptions are most effective when they correspond to our long-term interests. Thus, on non-ideal accounts one cannot demand that people continuously act against what they regard as being best for them in the long run (that they are "saints" or "heroes"). The clearer a moral prescription counteracts people's long-term interests, the weaker it is. For example, giving half of our wealth to the migrants and refugees of the world is something that morality just cannot demand on non-ideal accounts (Carens 1996, 161-163).
Finally, Carens discusses an "epistemological" precondition. From an epistemological point of view, he argues, morality is most effective if it corresponds to our local moral knowledge: to people's common beliefs about right and wrong. The assumption that states are widely free to admit or exclude aliens is supposed to be part of this knowledge. It is reflected both by our practices and principles. So on a non-ideal account there is no point in demanding to depart from this view (Carens 1996, 163-164).
Although Carens repeatedly emphasizes that non-ideal approaches do not rule out any criticism of the prevailing circumstances whatsoever, it is easy to see that the account he describes yields a morality that stands in stark contrast to his view in "Aliens and Citizens". If we take the above restrictions and preconditions into account, the conventional view turns out to be quite correct: states do not have a strong moral duty to admit immigrants.
Looking at things from an ideal perspective, however, Carens (1996, 169) still seems to believe that the conventional view turns out to be false and that borders should be more open than they currently are. After all, in this case the above restrictions and preconditions do not have to be taken into account. The question is not how migration should be regulated, given that the world is divided into nation states, given that people behave and think in certain ways, etc., but how migration should be regulated ideally, i.e., in an otherwise just world.
As pointed out above, in this paper Carens neither argues in favour of the non-ideal nor in favour of the ideal approach. He believes that both perspectives have their strengths and weaknesses. "Each approach has something important to contribute to our understanding of the ethics of migration. Yet I think that each approach typically brings one set of concerns into focus and simultaneously screens another from view" (Carens 1996, 157). The strength of the non-ideal approach is its relevancy. It is able to tell us what to do here and now. However, as explained, such an approach also tends to legitimize the current circumstances, even if they are (somewhat) unjust (Carens 1996, 164-165). The strength of the ideal approach is that it does not have this tendency. It is less tied to the current circumstances, and thus has more critical potential (Carens 1996, 166-167). Carens argues, however, that the idealistic approach suffers from a number of problems as well.
First, by taking an idealistic approach one's focus often shifts from the actual problem to very fundamental questions. One starts discussing the ethics of migration but ends up thinking about what a just world order would look like in generala question that is (a) largely independent from the ethics of migration and (b) very hard to answer (Carens 1996, 167). Second, even if one succeeded in figuring out what a just world order would look like this would not tell us how to get from our current nonideal circumstances to the ideal. This transition is not always straightforward. It is not necessarily most effective to change the actual circumstances in such a way that they are closer to the ideal (in an absolutely just world borders may be open; that does not mean, however, that the fastest or best way to open borders is starting to open borders here and now) (Carens 1996, 168). Third, by looking at migration from an ideal perspective a number of important problems disappear from view. For example, in an absolutely just world there would not be any refugees (refugees are per definitionem the product of unjust circumstances) (Carens 1996, 168).

Who Should Get in? The Ethics of Immigration Admissions
In "Who Should Get in? The Ethics of Immigration Admissions" Carens approaches the open borders debate from the perspective of extreme (or rather extreme) non-ideal theory. The non-ideal circumstances that are taken into consideration are mainly institutional and behavioural ones. Carens presupposes that the world is divided into sovereign and independent states and that these states have a "broad sovereign right to control immigration" (2003,95). Furthermore, he also grants the way in which states actually exercise this right.

Strikingly, and in contrast to what is suggested in "Realistic and Idealistic
Approaches to the Ethics of Migration", Carens comes to the conclusion that on such an approach the conventional view does not turn out to be right (2003,95,110). According to the conventional view, states' right to control immigration implies that the admission and exclusion of immigrants is not a moral issue. In fact, however, most states do treat immigration as a moral issue. Most importantly, almost all liberal democratic states acknowledge that they are morally obliged to admit immigrants in at least two cases: (1) in cases in which the immigrants are immediate family members of current citizens and residents, and (2) in the case of refugees.

Family Reunion
First, Carens points out that almost all liberal democratic states grant admission to current citizens' immediate family members, such as their children or spouses. This is true even for states that are known for strict immigration policies, and in many cases it is true not only for current citizens, but even for non-citizen residents (e.g., students, visiting professionals, or visiting workers) (Carens 2003, 96).
What is the reason for this generosity? Carens (2003, 96) argues that the reason is a felt moral obligation from the side of states and their representatives. States feel that they have a moral obligation to admit the immediate family members of their current citizens and residents. The obligation is felt not towards the outside family members, Carens (2003, 97) suggests. In this case it would be unclear why those people should be given priority over, e.g., people in extreme poverty. Rather, the obligation is supposed to be one towards the citizens or residents of the state itself. States consider themselves under an obligation to take into account the vital interests of their citizens and (in many cases) residents. Family reunion clearly exemplifies a vital interest. People's quality of life can be significantly impaired by being separated from their children, parents or siblings. Thus, most liberal democratic states ensure that citizens and residents can get their immediate family members to join them.
It might be objected that family reunions can take place not only within a given state, but outside of it as well. Each citizen and resident has a right to leave. So instead of his/her family joining him/her the citizen or resident could go to wherever his/her family currently lives as well (given that s/he will be admitted there). Does not this cast doubt on there being a moral right to family reunion? Carens responds by pointing out that citizens and residents do not only have a vital interest in family reunion, but a vital interest to continue living where they have lived as well. The state has to take this preference into account, and thus cannot shake off its obligation to admit the immediate family members of current citizens and residents by referring to the possibility of an external reunion (Carens 2003, 97).

Refugees
The second case in which most liberal democratic states acknowledge that they are under a moral obligation to admit immigrants is the case of refugees. One can distinguish between two kinds of refugees: (1) refugees that have been determined to be refugees by the UN High Commissioner for Refugees, and (2) asylum seekers, i.e., people who directly turn to the states they want to immigrate to and are selected by these countries (Carens 2003, 99-100).
In the case of refugees determined by the UN High Commissioner for Refugees, states normally do not consider themselves to be under a moral obligation to admission. Resettlement refugees have already found a safe place outside their home country. Thus, given states' broad right to control immigration, there does not seem to be a duty to admit them. States that do so (Carens mentions Canada and Sweden as countries that have admitted particularly many refugees in recent years, and the US as the leader in absolute numbers) certainly deserve praise. But states that do not, must not be blamed (Carens 2003, 100).
In the case of asylum seekers states' right to control immigration is restricted legally. According to the 1951 Geneva Convention Relating to the Status of Refugees and its 1967 protocol (signed by all European and North American states), people claiming to be refugees have a right to be heard by the state they want to immigrate to, and if their claim turns out to be correct, they have a right to stay. Carens (2003, 101) argues that this regulation is based on a moral obligation. In addition to their legal rights, asylum seekers also have a moral right to be heard, and if determined to be refugees, to stay. Unlike refugees, these people cannot simply be sent back to where they came from. In their home country they are in danger of being tortured or killed. According to Carens (2003, 102), states thus have a "deep moral obligation towards asylum seekers".
In addition to the cases of family reunion and refugees, Carens points out that, in some sense, states also treat ordinary cases of immigration as moral issues, i.e., cases in which the people who want to immigrate do not have any special rights to be admitted. States commonly consider themselves to be free to admit and exclude as many of these people as they like. However, they do not consider themselves to be free to do so on the basis of just any criteria. In particular, they believe that it would not be morally permissible for them to choose people on the basis of discriminatory criteria, giving preference, for example, to a certain sex, or a certain race (Carens 2003, 103-110).
One of the things Carens' article demonstrates quite clearly, and that was also emphasized in "Realistic and Idealistic Approaches to the Ethics of Migration" (1996,159), is that even a non-ideal approach allows for at least some degree of criticism. Carens does not approve of any aspect of the status quo. He acknowledges that liberal democratic states do sometimes determine admission and exclusion on the basis of discriminatory criteria. For example, he criticizes Germany's Aussiedler policy which gives preference to people of a certain ethnicity (2003,(109)(110). He also criticises some other practices regarding immigration, e.g., the US' giving higher priority to citizens than to non-citizen residents with regard to the admission of immediate family members (2003,98) or long waiting periods for the admission of children and spouses from some parts of the world in Canada (2003, 99).
This critical potential allows Carens to reject the conventional view even based on his non-ideal approach in "Who Should Get in? The Ethics of Immigration Admissions". According to the conventional view, states do not have a moral duty to admit immigrants. In fact, however, states behave as if they had such a duty. Although it is often politically unpopular and economically disadvantageous, they admit immediate family members and refugees and try to act morally correct in cases of ordinary immigration as well. This demonstrates, according to Carens, that the conventional view turns out to be wrong from a non-ideal perspective as well: The conventional view is that acceptance of the state's broad general right to control immigration means that morality has little role to play with regard to admissions. […] In practice, however, liberal democratic states do not treat their admissions decisions as morally unfettered.
[…] Even a minimalist account of the moral limits widely accepted by liberal democratic states imposes much greater restrictions on the states' discretion with regard to immigration than the conventional view allows. (Carens 2003, 95)

Concluding remarks
In the past 30 years Joseph Carens' contributions to the open borders debate have gradually taken on a different complexion. Starting at an extremely ideal level of analysis, Carens has become more and more concerned about what we ought to do under current non-ideal circumstances. The three articles analyzed in this paper reflect this shift quite clearly.
In "Aliens and Citizens: The Case for Open Borders" (1987) Carens looks at the open borders debate from the perspective of ideal theory. He is interested in how immigration ought to be regulated in a perfectly just world. In "Realistic and Idealistic Approaches to the Ethics of Migration" (1996), both the idealistic and the non-idealistic approach are discussed, with Carens suggesting that both perspectives are important, and that they complement rather than exclude each other. In "Who Should Get in? The Ethics of Immigration Admissions" (2003), finally, Carens examines the open borders debate at an extremely non-ideal level of analysis.
The common thread that runs through all of these works is Carens' opposition to the conventional view regarding immigration (the view that liberal states do not have a strong duty to admit immigrants; or at least, they do not have a duty to admit all immigrants). Understandably, this opposition manifests itself most clearly in "Aliens and Citizens", where empirical realities are hardly taken into consideration at all. However, contrary to what is predicted in "Realistic and Idealistic Approaches to the Ethics of Migration", the opposition is even present in the non-ideal analysis of "Who Should Get in?". The aim of this paper was mainly illustrative. In closing, however, let me also make two brief critical remarks: one that pertains to Carens' critique of the conventional view, and one that pertains to ethical debates about open borders more generally.
First, it seems to me that some of Carens' arguments rest on a slight misrepresentation of their target, i.e., the conventional view. He suggests that the conventional view does not only attribute little role to morality with regard to admissions, but almost, or even literally, none. In "Aliens and Citizens", for example, the conventional view is introduced as the view that states "[…] may choose to be generous in admitting immigrants, but they are under no [sic!] obligation to do so" (1987,251). But only few philosophers have held the conventional view in such an extreme form. In particular, this is not how the view was defended by the philosopher who has been Carens' main target, namely Michael Walzer. In "Spheres of Justice" (1984, 41) Walzer explicitly notes that states can be compared to families, and that they thus have at least some moral obligation to admit family members of current citizens and, in some cases, displaced ethnic nationals. Moreover, Walzer (1984, 33) concedes that there is a moral obligation to admit refugees seeking asylum (Wilcox 2009, 2-3).
Second, while my focus in this paper was on Carens, its conclusions are supposed to apply more widely. In the end any philosopher's stance on the debate about open borders (and the ethics of migration more generally) is influenced by the ideality of his or her approach. This methodological insight has not yet received sufficient attention. While Carens himself is admiringly aware of it, other participants of the debate have (largely) failed to realize that their claims are contingent on the extent to which they account for empirical facts. Hopefully, my considerations show that the fundamental question of how ideal or non-ideal one's approach should be cannot simply be bracketed. Discussions about liberal states' duties with regard to immigration will benefit much from making underlying idealityassumptions explicit, and from assessing them in ways similar to those employed by Carens.