Human rights begin … [i]n small places, close to home …. Unless these rights have meaning there, they have little meaning anywhere. Without concerted citizen action to uphold them close to home, we shall look in vain for progress in the larger world.

— Eleanor Roosevelt

There has been extensive philosophical debate about the requirements of international justice. Yet virtually anyone would agree that minimally international justice requires the protection of asylum-seekers, as outlined by international conventions and treaties. Identifying agents capable of discharging such international obligations arising from international law becomes thus crucial (O’Neill 2001).

States and their national governments have commonly been regarded as agents of international justice; agents who can promote compliance with the aforementioned international obligations (O’Neill 2001; Pogge 2002). International organizations (IOs) and even non-governmental organizations (NGOs) also play an important role, especially when State capacity is low or inexistent (e.g., Keck and Sikkink 1998; Rubenstein 2015). This article focuses on yet another category of agents that can promote international justice by ensuring compliance with international law: subnational units (cities, states, and provinces in federal States, autonomous regions, and municipalities in non-federal ones).Footnote 1 Its aim is to examine how subnational units can advance international justice by promoting the respect of international law.

Drawing on legal discussions that normative political theorists largely neglect and taking the USA as its primary case study, the article proposes three theoretical models for understanding the responsibility subnational units have as well as the strategies they can pursue to promote international law. I discuss what these responsibilities and strategies entail in the case of one area of international law: refugee protection. I understand refugee protection to require at a minimum non-refoulement and the fulfillment of the most basic human rights for refugees.Footnote 2 The implementation of human rights treaties is thus crucial for refugee protection.

The significant autonomy that self-determining subnational units enjoy in many areas—such as law enforcement, education, housing, healthcare, and social welfare—makes such units capable of fulfilling international obligations in the area of migration. Morally as well, local communities should have a say in migration policy insofar as the costs and benefits of receiving and integrating migrants are felt primarily at the local level. Local institutions, such as state legislatures or municipal councils, enable local communities to make binding collective decisions in matters that concern them; these matters sometimes include refugee protection, as I point out here.

The article builds on existing normative arguments about the role of subnational units. Normative political theorists already generally assume that all individuals and groups (subnational units included) have moral duties to protect asylum-seekers (see, e.g., Nickel 1993; Alston 2005; Brock 2020). Much less discussed and theorized, however, are the legal mechanisms by which they (and particularly subnational units) can do so.Footnote 3 Looking at the real-world constraints that block or hinder the performance of such duties, as well as the legal opportunity structures that create possibilities for discharging them, is thus an equally important task. In systematizing the various mechanisms that allow subnational units to promote compliance with international law, this discussion aims to ultimately help political theorists better anchor their normative arguments in present-day legal realities.

The article conceptually distinguishes three models for understanding subnational units’ relationship to international law: (1) derived delegated responsibilities; (2) derived back-up responsibilities; and (3) assumed responsibilities. The first two types of responsibilities pertain primarily to international obligations incurred as a result of treaties signed and ratified by the national level. The third applies to treaties that have not been signed, or have not been ratified (or have been ratified subject to major reservations). This model could in principle extend to other non-treaty-based international obligations, such as those imposed by customary law, but those will not be discussed here.

In discharging derived responsibilities, subnational units can employ either a positive or a negative strategy for doing so. Under the positive one strategy, subnational units may adopt local or provincial legislation consistent with international law or become parties to international agreements (if not treaties) aiming to promote it. Under the negative strategy, subnational units may refuse to enforce any orders coming from the national level that clash with the country’s existing international commitments. Sanctuary for asylum-seekers, a practice that has received much press attention over the last years, is a good example.

While the discussion will focus primarily on international commitments to protect asylum-seekers by upholding non-refoulement and respecting their basic human rights, the analysis put forward here could extend to international commitments in other areas as well. Climate change and environmental protection, areas where subnational units have been actively involved of late, also serve as a good illustration of the different types of responsibilities and mechanisms that enable subnational units to advance international law.

Although having far wider application, the theoretical models of responsibility proposed here are inspired and exemplified by the USA, whose complex federalist structure interestingly acts as a two-edged sword in the area of refugee protection. This complexity makes the USA a particularly interesting but also challenging case. This case is worth special attention also because, as the most important world power, the USA can set an example for other federal States around the world through its own record.

Subnational Capacity: Reserved Powers

Because refugee protection falls in the area of international affairs, it is worth having a brief look at the legal mechanisms that may allow (or prevent) subnational units to play any role in this area. All three approaches to subnational protection of refugees that I shall be discussing hinge on subnational units having the legal capacity to do so. In the US system that legal capacity derives from the constitutional sources that I shall be discussing in this section.

The US states enjoy exclusive reserved powers in many domains. As we will see in the “Derived Responsibilities: the Delegated Responsibility Model” and “Derived Responsibilities: the Back-up Responsibility Model” sections, these exclusive powers enable subnational units to pursue both negative (sanctuary) and positive strategies to protect refugees. Effectively, according to the Tenth Amendment, all powers that have not been expressly delegated by the US constitution to the federal level, nor prohibited by it to the states, belong to the states. The Tenth Amendment ensures thus that the states retain the residuum of unenumerated powers, while Congress and the President can exercise only the powers explicitly vested in them by the constitution (Glennon and Sloane 2016, pp. 16–21). Similarly, the states’ own constitutions might provide exclusive powers for cities such that they retain some autonomy vis-à-vis the state they are a part of.

Foreign affairs powers are mainly vested in Congress and the President, and states are clearly prohibited from doing certain things in the area of foreign affairs.Footnote 4 Importantly, however, the US constitution does not prohibit the states from taking any action that might affect foreign affairs. Nor does it clearly state that the federal government’s foreign affairs powers are exclusive (Glennon and Sloane 2016, chs. 2, 4; Strauss 2014, p. 426; Henkin 1996, pp. 150–1). Furthermore, it does not say what states may do “until or unless” the federal government prohibits certain activities through treaties (Glennon and Sloane 2016, p. 88). This leaves plenty of room for states to act when the federal level has not adopted any clear policy, ratified any relevant treaty or adopted any implementing legislation for a ratified, non-self-executing treaty.

With the expansion of the scope of international treaties (human rights treaties being a case in point), the foreign affairs powers of the federal level are increasingly intersecting with the reserved powers of states. At the same time, subnational units have and are increasingly using their own powers in areas that may directly or obliquely impact on foreign affairs (Glennon and Sloane 2016, ch. 2).

The US states’ reserved powers matter crucially to their capacity to promote or thwart international law. Thus, in principle, a treaty may be deemed unconstitutional and may fail to preempt state law if it encroaches on the state’s Tenth Amendment powers (Glennon and Sloane 2016, pp. 190–4, 308–9). Implementing legislation adopted by Congress to make a non-self-executing treaty enforceable by courts may similarly be deemed unconstitutional if it interferes with the states’ reserved powers. And states can play some role in foreign affairs as long as state law does not conflict with—and thus is not preempted by—federal law including international treaties. At the same time, the Tenth Amendment limits the federal government’s capacity in some areas of international affairs, like immigration, as discussed in the “Derived Responsibilities: the Back-Up Responsibility Model” section.

Some states sought to use their powers to restrict or afford undocumented migrants access to education or employment (Glennon and Sloane 2016, p. 73). This is unsurprising. Immigration in particular illustrates the complexity that federalism creates in the area of international affairs. While the “federal government enjoys potentially plenary power .... it has not, expressly or implicitly, ‘occupied the field’ to preempt state regulation of immigration. Nor, as a practical matter, could it probably do so. The simple fact is that the federal government lacks sufficient resources—financial, temporal, law-enforcement personnel, and so forth—to handle immigration entirely on its own .... federal government has left major aspects of immigration regulation to the states” (Glennon and Sloane 2016, p. 301). Importantly, while the federal government has significant powers in the area of immigration, its capacity to enforce federal laws and policies can be thwarted by states, as we will see below.

Generally, state laws cannot be preempted if the federal level has not taken any legal action on a matter.Footnote 5 One exception is the dormant foreign affairs doctrine (introduced by the Supreme Court in Zschernig v. Miller) according to which the federal level can preempt state law in foreign affairs even in the absence of any federal law. Yet, support for this doctrine has been feeble over time (Glennon and Sloane 2016, ch. 2; Strauss 2014, p. 428), and courts have most often sided with states (e.g., Blythe v. Hinckley, Clark v. Allen; Barclays Bank PLC v. Franchise Tax Board of California) claiming that if there is no relevant treaty or policy, states retain the right to legislate on matters affecting foreign affairs using their reserved powers.Footnote 6 In 2003, Zschernig was invoked again in a case that appeared to set new limits to states’ capacity of acting in the area of international affairs. A Californian law was struck down by the Supreme Court in the absence of any federal law or treaty and merely on the basis of a conflict with an executive policy interest (Strauss 2014, p. 434).Footnote 7 The scope of conflict preemption has thus been expanded.

The decision has been widely debated and criticized on multiple grounds (e.g., Glennon and Sloane 2016, pp. 129–46).Footnote 8 Debates over its strength and standing are beyond the scope of this article and orthogonal to its purpose. Despite their ever-changing balance, international affairs are bound to clash with federalism (Strauss 2014, p. 420). But the mixed bag of court decisions shows that while over the years states have successfully legislated in the area of foreign affairs, in the future, any given decision by a state is vulnerable to being struck down by the Supreme Court. But this may be unlikely considering the long-time record of Supreme Court decisions (mostly against the dormant foreign affairs doctrine), as well as the Court’s decision to defend states’ police powers in other iconic cases like Medellín.Footnote 9 Furthermore, while Garamendi has expanded conflict preemption, it has also narrowed the dormant foreign affairs doctrine stating that it cannot be invoked if states used their traditional police powers under the Tenth Amendment and the federal government has not acted.Footnote 10 This means that, if there is a gap in foreign policy, states are actually welcome to fill it as long as they are using their exclusive, traditional state police powers (Strauss 2014, p. 452). This position is fairly similar to that enjoyed by subnational units in other federal states like Switzerland and Canada (Glennon and Sloane 2016, pp. 362–6).

In a nutshell, the states’ police powers are what allows them to promote international law. As we will see below, they enable these subnational units to play a role in refugee protection through either a negative (sanctuary) or a positive strategy.

The Derived Responsibilities of Subnational Units

What may justify subnational units promoting compliance with international law, where they have the capacity to do so?

Insofar as international law binds a State, it also indirectly legally binds all of its subnational units, even if the national level is the only one that is accountable internationally. This is an argument for thinking that international responsibilities do trickle down to the subnational level and that subnational units have, as members of the State, derived duties to act on them.

While State practice might vary (in some States international treaties have the same status as the Constitution, in others they are superior to it (Mendez 2013, ch.1)), within the US domestic legal system, international law has the status of federal law. Just like federal law, it is superior to domestic state law and thus binding on subnational units. The Supremacy Clause of Article VI of the Constitution clearly subordinates subnational, state law to the US’ treaty obligations; “all treaties shall be the Supreme Law of the Land; and the judges in every state shall be bound thereby, anything in the Constitution or Laws of any state to the contrary notwithstanding” (US Constitution, art. 6, §2. See further Tribe 1988, p. 226).Footnote 11, Footnote 12 This means that the states cannot adopt legislation contrary to treaty obligations. Indeed, as Federalist no. 22 and 42 explain, by ensuring that treaty obligations will override conflicting state laws in state courts, the Founding Fathers wanted to prevent states from violating treaty obligations (American Law Institute 2018, p. 75; Kaufman 2012, p. 119). There is thus a clear sense in which subnational units are legally bound, at least indirectly, by the international commitments undertaken by the federal level.Footnote 13

We may also argue that subnational units are also bound by treaties, politically and legally more generally, as members of the collective agent that is the USA. Treaties are negotiated and concluded the president; the Senate must then give its advice and consent to their ratification by the president. As part of a political union, the states must honor the obligations that were undertaken by the union. Any obligation incurred by a collective agent (the union), thereby, trickles down to bind its individual member states. This view is clearly articulated in the Supreme Court’s assertion that when it comes to international negotiations and compacts, all states speak with one voice, that of the USA (United States v. Belmont, 301, U.S 324, 331 (1937). The US Congress has also clearly declared that “states have an obligation through the federal system to implement international human rights obligations undertaken at the federal level.”Footnote 14

Furthermore, the president can ratify treaties only after they are given the advice and consent of the Senate, assembling elected representatives from all constituent states. The consent of the US states is crucial to whether the country incurs an international obligation to obey a treaty or not. Most states, through their political representatives, must have clearly committed to those treaty obligations when the Senate passed a resolution of consent to ratification of the treaty. In virtue of this process of democratic authorization, we may also argue that ratified treaties procedurally bind those states.

All constituent states have, of course, also committed to the US Constitution—and the Supremacy Clause contained within it which provides that treaties trump state law—when they ratified the Constitution. This means the states have also consented to the status of treaties as federal law.

Cities and municipalities were not directly involved in these processes, to be sure. But as “creatures of their states,” they are also derivatively bound by the commitments of the states of which they are subsidiaries.Footnote 15 And the same would, of course, be true of municipalities in unitary States. We can thus reasonably suppose that subnational units do share in the national level’s international commitments. Politically and legally, we could argue that they are bound by them.

Derived Responsibilities: the Delegated Responsibility Model

Above, I argued that subnational units have derived responsibilities to promote the international obligations that the State has undertaken through ratified treaties. We can think of two models for understanding those derived responsibilities: delegated and back-up responsibilities.

If a treaty is non-self-executing, then, by definition, further implementing legislation is required to give it domestic legal effect. A State ratifying a non-self-executing treaty commits to implement that treaty domestically. But if under that State’s constitution such legislation unavoidably falls within the “reserved powers” of subnational units, then the State’s constitution dictates that only subnational units can adopt the implementing legislation required to meet the State’s treaty obligation. While the subnational units’ responsibility to act derives from the State’s treaty obligation, it is of a delegated kind. The delegation in question comes not from any specific act of Congress but rather from the State constitution, which reserves for the subnational units certain powers that may be needed for treaty implementation. Thus, the delegation of subnational units to act in some area comes from the constitution, while the mandate to act comes from the State’s ratification of some international treaty mandating such action.

In the case of the US states, this framework of interpretation is not unfit for understanding the responsibilities that they bear in the area of refugee protection. Protecting refugees requires not just non-refoulement but ensuring that their most basic human rights will be fulfilled. Yet, many human rights treaties that have been ratified by the USA are non-self-executing. While human rights law is meant to create entitlements for individuals, if a human right treaty is non-self-executing and no executing, implementing legislation has been adopted, then courts cannot use these treaties as decision rules. In other words, the treaties are not enforceable by courts as they create no private rights of action (American Law Institute 2018, p. 75).

Many human rights treaties have not been executed by the US Congress in an explicit deference to federalism. Even when it has agreed to a treaty’s ratification, the US SenateFootnote 16 has sometimes declined to adopt any federal implementing legislation invoking the exclusive, traditional powers of states (Kaufman 2012, p. 120).Footnote 17 Importantly, both executive and legislative branches have encouraged the local implementation of human rights treaties (Davis 2008, p. 412). Thus, the Child Support Convention has been largely implemented through the adoption of a uniform state law drafted with the help of the Uniform Law Commission, which was then adopted by every US state (Ku et al. 2019, p. 130).

In ratifying the International Covenant on Civil and Political Rights (ICCPR), the Senate also clearly declared that the states independently control the implementation of these treaty obligations (Ku 2004, p. 521; Kaufman 2012, p. 119). The ratification of the Convention against Torture and Other Cruel, Inhuman, or Degrading Treatment or Punishment (CAT) and the Convention on the Elimination of All Forms of Racial Discrimination (CERD) contained similar acknowledgments (Ku 2004, pp. 521–3, 525).Footnote 18 Meanwhile with respect to the ICCPR, the president also explicitly pointed out that the federal government will “remove any federal prohibition to the abilities of the constituent states to meet their obligations in this regard” (Kaufman 2012, p. 120, emphasis added).

The federal government has also called for local action in the Medellín case, pointing out to the International Court of Justice that, because the USA is a federal republic, the federal government “does not have the legal power to stop” the execution of foreign nationals even if it would put the USA in breach of its international obligations (Ku 2004, p. 462).Footnote 19

While state legislatures have in particular been singled out as bearing responsibilities in the area of treaty implementation, some scholars argue that state governors should also shoulder these responsibilities: “when the federal government ratifies a treaty or declares adherence to a norm of customary law that implicates state functions or interests, state governors are responsible for implementing those treaty obligations” (Ku 2004, p. 461).

To be sure, at least some types of non-self-executing treaties may be implemented in different, alternative ways: they could either be implemented through federal legislation or through local law by subnational units. In such a case, where the implementation of a treaty does not uniquely depend on the subnational units’ reserved powers, the derived responsibilities of subnational units cannot be thought as being of a “delegated” sort. Instead, they might be better thought of as being “back-up” responsibilities (discussed next) that arise only when the federal level has failed to act on its treaty commitments and adopt federal implementing legislation. In that case, states can move in and fill the gap in implementation, but they would do so in response to the failure of Congress to execute the treaty when it can do so, rather than because the states are the primary bearers of responsibility for implementing those treaties.Footnote 20 Instead, if the state has a delegated responsibility—that is, a treaty can be executed only through local implementing legislation adopted by the states in virtue of their reserved powers—then this responsibility kicks in immediately upon the ratification of a treaty, provided that the federal government has failed to act on its treaty commitments.

Subnational units having derived delegated responsibilities puts Congress’ own inaction in a different light. It denotes less a failure or refusal of the federal level to act on its international commitments, and more the fact that the states, not the federal level, should be the first bearers of responsibility for treaty implementation.

Reasons for adopting such a distribution of legal competences might be found in the distribution of practical competences. Often it is best to implement treaties at the lowest possible level. A principle of subsidiarity may thus support local treaty implementation (Aust 2020). Human rights treaties are a good example.Footnote 21 For instance, social services by their nature are better delivered by the level of government that is “closest” to the potential recipients. The rights to health or education of refugees also depend heavily on ensuring their access to the local services of the state or city sheltering them.

Derived Responsibilities: the Back-Up Responsibility Model

The second model of derived responsibility for subnational units is one of back-up responsibility. If we understand subnational units’ derived responsibilities as of a back-up sort, then they kick in only because the federal level has failed to act on its international commitments when it had capacity to do so constitutionally. Then, subnational units have a back-up duty to act so as to ensure that the effects of these failures are mitigated at the local level. Depending on the source of the failure—either executive orders that clash with these obligations or a failure of Congress to execute (i.e., pass implementing legislation for) non-self-executing treaties—subnational units can adopt either the negative (sanctuary) or the positive strategy (domestic incorporation) discussed below.

In contrast to the previous model of delegated responsibility, where subnational units would have an immediate, primary duty to domestically incorporate a treaty through implementing legislation, under the back-up model they would do so only when the said treaty could have been implemented otherwise by Congress but Congress failed to do so.

A. The Negative Strategy: Sanctuary

If the federal government adopts any policy in breach of international law, subnational units can refuse to enforce such regulations locally. Just as soldiers can refuse to obey their superiors’ orders if they contravene international humanitarian law, so too, we might argue, subnational units can use their exclusive powers to circumvent orders from the federal level that contravene treaty-based international obligations that have not been intentionally and explicitly repudiated by Congress.

Sanctuary offered to asylum-seekers is a good illustration of this negative strategy for discharging derived back-up responsibilities in the area of refugee protection.Footnote 22 By refusing to collaborate with federal agencies to deport asylum-seekers, within the powers assigned to them constitutionally, subnational can prevent the deportation of asylum-seekers to dangerous places that would constitute refoulement. Sanctuary cities and sanctuary states can thus effectively act as spaces devoted to the promotion of international law.

Many cities declared themselves sanctuaries in a push-back against President Trump’s anti-migrant statements and policies. Chicago, Philadelphia, and Seattle, through their mayors, declared themselves ready to take in Syrian refugees despite the national government’s reluctance to do so (Vyse 2019). And while some US states have, with more or less success, forbidden sanctuary cities within their territory,Footnote 23 others, like California, Colorado, Illinois, Massachusetts, New Mexico, New Jersey, Oregon, and Vermont, claim the status of “sanctuary” for their entire territory. Outside of these eight sanctuary states, 87 counties or municipalities claim sanctuary status (Hudak et al. 2019).

Generally, sanctuary cities or states are jurisdictions that enjoy significant territorially-specific discretion in enforcing federal law. This means they can rely on different legal instruments to escape any orders from the higher level that might conflict with international law. In the USA, states and municipalities might avail themselves of two different legal doctrines to pursue sanctuary policies.

A1. Anti-commandeering

States may refuse to enforce federal orders on the ground that such orders violate the anti-commandeering doctrine.Footnote 24 Rooted in the Tenth Amendment, the doctrine holds that “‘even where Congress has the authority under the Constitution to pass laws requiring or prohibiting certain acts, it lacks the power directly to compel the states to require or prohibit those acts’ on behalf of the federal government” (Santamaria 2020). Thus even if the federal government may preempt state and local laws through its legislation, federalism prevents the national level from “commandeering” the resources of the states to enforce national laws and policies on the ground.Footnote 25 Thus, the anti-sanctuary orders of the Trump administration that demanded or prohibited particular actions from states’ officials in order to enforce immigration law (or conditioned local funding on compliance with these orders) were deemed unconstitutional by many courts.Footnote 26

A2. Home Rule

Municipalities in general are legally “creatures” of the state and operate under the authority granted by it. As such, municipalities too can appeal to the anti-commandeering doctrine to resist the anti-sanctuary efforts of the federal government (see Gulasekaram et al. 2019, p. 847 citing Country of Santa Clara v. Trump, 275 F. Supp. 3d 1196, 1215–16; City of Philadelphia, 280 F. Supp at 651; City of Chicago, 264 F. Supp. 3d at 949). If, however, that avenue is blocked because their state has adopted its own anti-sanctuary legislation (as Alabama, Mississippi, North Carolina, Indiana, Tennessee, Iowa have in the past),Footnote 27 the municipalities may nonetheless appeal to a different legal doctrine: the doctrine of (municipal) home rule, which structures the municipalities’ relationship to their states.Footnote 28

The power of states over their localities may also be limited by the states’ own constitutions or laws. The autonomy municipalities can enjoy under home rule will vary.Footnote 29 But when it comes to immigration policy, which relies heavily on local participation in enforcement, home rule could limit states’ capacity to impose anti-sanctuary policies on their municipalities.Footnote 30

B. The Positive Strategy: Domestic Legal Incorporation

If the source of the failure to act on international commitments stems from Congress’s failure to adopt implementing legislation for non-self-executing treaties when it can do so (i.e., when a treaty could be implemented without resorting to the powers reserved for the states under the Tenth Amendment—see my discussion in the “Derived Responsibilities: the Delegated Responsibility Model” section), subnational units may adopt a positive strategy to discharge their back-up responsibility. State legislatures could then adopt implementing legislation, incorporating ratified treaties into domestic law and thereby giving them full effect in the states’ courts, even when Congress has declined to do so.Footnote 31 And cities could incorporate treaty commitments into their charters and codes of practice of their local services.Footnote 32 This is precisely how some legal scholars envisage the responsibilities of subnational units: “where Congress has failed to take this action [passing implementing legislation] (or believes it unnecessary), the states are responsible for carrying out those treaty obligations, usually through legislation” (Ku 2004, p. 462). It is thus primarily through treaty implementation that subnational units’ derived international responsibilities come to the fore.

Of course, where it would be within its powers to do so, one would first look to the national legislature for the adoption of implementing legislation. Implementing legislation passed by Congress would preempt any legislation that would adopted by the states, thereby preventing them from acting on their derived obligations.Footnote 33 However, the USA and other federal countries have often declined to take the lead. They entered reservations to treaties in order to encourage subnational units to play some role in treaty implementation. Other times, some leeway was explicitly included in the federal legislation that Congress adopted.

Cities can also follow a positive strategy in the area of refugee protection. Many municipalities have incorporated the Universal Declaration of Human Rights into their local laws. This is the case of New York, Chicago, San Francisco, Cincinnati, Chapel Hill, Madison, Portland, Los Angeles, Berkeley, and Washington (Kaufman 2012, pp. 128–35). In other cases, principles underlying human rights law were incorporated into local charters, just without any explicit reference to those treaties (Gonzalez 2016).

There are numerous cases of subnational implementation of treaty obligations. As a result, subnational implementation has been touted as one good way of closing the gap between the “normative” and the “operating” systems of international law—that is, between its formal, prescriptive commitments, and their realization (Ku et al. 2019, pp. 108, 117). Take for example, consular protection and the now famous Medellín case.Footnote 34 While the International Court of Justice (ICJ) deemed the USA in breach of Article 36 of the Vienna Convention on Consular Relations, the Supreme Court argued that the state of Texas was not obliged to enforce the ICJ’s judgements since such judgements are not federal law. Yet this enforcement gap of international obligations created by the Supreme Court’s decision, which could in principle be rectified by Congress, was tackled locally by some US states. California, Illinois, and Oregon chose to implement Article 36 through state or local law (Ku et al. 2019, p. 111).

In light of this, some scholars argue that the US states in fact “control compliance with international law,” fulfilling “US responsibilities under both customary international law and treaties” (Ku 2004. p. 461).Footnote 35 For them, the states are not just obliged to obey international law qua federal law. Instead, they say, current practice shows that the states “feel obligated to obey international law as ‘international law’” (Ku 2004, pp. 498, 525). Some judges have endorsed this view as well. For example, one Oklahoma judge declared in a concurring opinion that that state has “an independent international obligation” to respect the rights of a prisoner under the Vienna Convention (Davis 2008, pp. 419–20, emphasis added; see also American Law Institute 2018, pp. 79–80).

The theoretical framework developed here for understanding subnational units’ responsibilities may to some extent support the view that states “control compliance” with treaties. If they have delegated responsibilities, a model that could apply at least to some non-self-executing treaties, then we could say that states control compliance insofar as the treaty can only be implemented domestically through the action of subnational units. On the other hand, the second model of back-up responsibility also supports this assessment insofar as, on this model, the states are the last resort for ensuring compliance with treaties when Congress has failed to do so by its own means.

Assumed Responsibilities: Toward a More Independent Role for Subnational Units

Above, I discussed how subnational units can promote international law by acting according to either a delegated or a back-up model of derived responsibility. As a reminder, derived responsibilities, I argued, exist vis-à-vis treaties that have been signed and ratified (The Derived Responsibilities of Subnational Units section). But even when the federal level has not signed and ratified (or has ratified with substantial reservations) and thus is not (fully) committed to any given international treaty or convention, subnational units may still be able to themselves adhere to those international norms within their own jurisdictions.Footnote 36 In other words, they can promote international law also by acting according to a model of assumed responsibility. This is the third model of subnational unit responsibility to which I now turn.

There are various ways in which subnational units can promote international law by assuming responsibilities for unratified treaties, treaties that have not been signed, or treaties that have been ratified subject to extensive reservations. As we saw in the “Subnational Capacity: Reserved Powers” section, provided there is no federal law or policy that may clash with local law, provided states are using their traditional police powers, and provided the effect on foreign affairs is minimal, they can effectively legislate in the area of international affairs.

While they cannot legally join a treaty or convention as signatory parties (the treaty power being reserved for the federal government), subnational units in the USA (and elsewhere) can nonetheless incorporate into their own state and municipal law treaties that have not been ratified or have not even been signed. They may even negotiate their own agreements with other foreign governments, regions, or cities. This is the positive strategy they can pursue to discharge assumed responsibilities. Alternatively (or in addition to it), subnational units can pursue a negative strategy: they may circumvent (within their own jurisdictions) treaty reservations, incorporating into their own law the norms the Senate has explicitly refused to commit to.

The USA has refused to ratify several human rights treaties pointing to the limits imposed by federalism (Burroughs 2006, pp. 412–3). This is the case, for example, of the Convention on the Elimination of All Forms of Discrimination Against Women (CEDAW) and the Convention on the Rights of Persons with Disabilities. Despite being signed in 1995, the Convention on the Rights of the Child has not even been sent to the Senate, making the USA the only country in the world not to have ratified it—largely due to the constraints of federalism (Ku et al. 2019, p. 112). Despite this, the gap in human rights protection has been closed through local action.

Thus, in pursuing the positive strategy, some states have incorporated these human rights norms into their domestic law and constitutions. For example, article 1 of the California Constitution partially incorporates the International Convention on the Elimination of All Forms of Racial Discrimination (US Human Rights Network’s CERD Working Groups 2008, pp. 17–8), a convention that the USA has ratified only subject to great many reservations (Watson 2020). The California Senate and Assembly even passed legislation requiring state and city officials to prepare periodic reports for the U.N Committees overseeing the treaty (Kaufman 2012, p. 130).Footnote 37 Hawaii, Rhode Island and South Carolina have, on the other hand, endorsed the principles of the Convention on the Rights of the Child (Ku et al. 2019, p. 113).

Similarly, municipalities can incorporate unratified treaties into their municipal law. San Francisco has, for example, famously adopted the Convention on the Elimination of All Forms of Discrimination against Women (CEDAW) through a municipal ordinance in 1998 (US Human Rights Network’s CERD Working Groups 2008, p. 13). Other cities have joined the Cities for CEDAW initiative that aims to incorporate principles of gender equity and the obligations of CEDAW into city governance and local policy (Ku et al. 2019, p. 113). Many other cities have either passed CEDAW ordinances or CEDAW resolutions affirming support for CEDAW principles.Footnote 38 Other unratified treaties received a similar treatment. Austin, Chicago, and Savannah embraced the principles of the Convention on the Rights of the Child (Ku et al. 2019, p. 113).Footnote 39

The subnational implementation of unratified treaties is also a way of putting pressure on the Senate to ratify a certain treaty in the future. The Cities for CEDAW initiative aimed to do just that (Powell 2001, p. 279). The strategy has yielded results in the past. Take, for example, the Hague Convention on the International Recovery of Child Support and other Forms of Family Maintenance.Footnote 40 It was ratified in 2016 only after each US state adopted a common set of standards complying with the substantive obligations of that Convention, essentially implementing them locally (Ku et al. 2019, pp. 139–41).

The Future of Local Internationalism

As we saw, subnational units already have means to effectively promote international law. In light of these legal instruments, it would not be unreasonable for political theorists to focus more on how local agents (as opposed to nation States) can advance cosmopolitan agendas. This section completes our foray into the question of how subnational units can be active agents in the area of refugee protection by upholding international law. It briefly entertains two reform proposals: (1) an expansion of the powers that subnational units enjoy and (2) making subnational units formal makers as well as takers of international law. In the future an increased role for subnational units might be a good way of bridging the justice gaps that arise from the failure of national governments to act.

A. Proposal 1: Extended Powers

While there is much that subnational units can already do, their ability to promote human rights has limits. The reason is that some important powers are exclusively in the hands of the national government. For example, the US states and cities do not have power to control national borders. Even if a state like California “opens” its borders to asylum-seekers, asylum-seekers may not be able to take advantage of this if US border agents block them from entering the USA and arriving in California.

Of course, we may argue that there are good reasons why subnational units should have a say in federal immigration policy. If subnational units are willing to take in refugees and other migrants, they should be allowed to do so. After all, the local communities where asylum-seekers settle will be the most affected by their presence; in virtue of this, these communities deserve to have a say on whether such people are allowed to enter and stay in their state or city. Immigrants will also require access to healthcare and education, which will often be provided through local income, sales, and property taxation. If subnational units are prepared to take in these people and bear the costs of their protection, then the national government has good reason to allow these people access, at least on condition that they do not move elsewhere without the agreement of the other subnational unit to which they are moving. In order words, we could argue that subnational units should be allowed to “vouch” for these migrants even if the national government is opposed to their presence. While some provinces and cities might be unwilling to open their doors, at least such a policy would allow those provinces and cities that want to welcome refugees to do so. This would be preferable to a situation where the national government closes its borders completely to asylum-seekers or takes in only a small number of them (smaller than the number of migrants that would be accommodated through the alternative proposal where subnational units have a say in these matters).Footnote 41

A1. The Canadian Model

Canada has recognized something similar through its program of private sponsorship of refugees.Footnote 42 The Canadian government admitted a substantial number of refugees but also allowed Canadian citizens—alone, in partnerships, or through associations—to sponsor additional numbers of refugees. The citizens, partnerships, and local associations sponsoring these refugees are responsible for their integration. They must offer social, residential, and financial support, as well as food and clothing (Immigration, Refugees, and Citizenship Canada 2017). A similar program could be adopted (in the USA but also elsewhere) by which subnational units can apply to the federal government and request that they be granted the right to receive and protect a certain number of refugees.

B. Proposal 2: Formally Recognized International Responsibilities

The option of formally acknowledging subnational units as makers and takers of international law is more fanciful, and perhaps less palatable. After all, the standard view within international law has always been that States, and only States alone, can be makers and takers of (i.e., held accountable by) such law.

This classic view does not however completely stands the empirical test. Not only has international law occasionally recognized cities as legal persons without recognizing them as States (The International City of Tangiers and the Free City of Danzig are well-known examples).Footnote 43 In some areas, international law is also already creeping in to regulate the activity of subnational units (Frug and Barron 2006; Blank 2006). It is doing so by imposing constrains on the exclusive local powers that subnational units rightly enjoy, as if these units (not just the State they are a part of) were formal subjects of international law (Frug and Barron 2006, p. 21).

The decisions of international arbitration tribunals in particular have directly impacted urban government in areas that have commonly been under subnational units’ exclusive control: local land use and waste treatment (Frug and Barron 2006, pp. 40–51). International trade and investment agreements are also increasingly encroaching on subnational units’ powers, especially those pertaining to their relationships to foreign investors (ibid., pp. 36, 38). What these decisions highlight is that subnational units are in fact bound by international regulations and denied certain actions even when they are compliant with their State’s federal and constitutional law. Subnational units themselves have become more aware that they are “increasingly obligated under trade rules and policies,” as the California Senate itself remarked (ibid., p. 39). There is thus a clear sense in which subnational units are already treated by the international system as “takers” of international law, having their local powers subjected to the decisions of international arbitration courts.

At present, subnational units’ liability under international law is of course wholly derivative from their national governments’—not their own—direct liability under international law. There is still some way to go before international law formally asserts full authority over subnational units directly, as subjects of international law in their own right. Despite this, some states and cities are already trying to assert such a status while being active players in global governance.

California, for example, has already independently joined United Nations’ efforts in some areas, irrespective of the federal level’s commitment. It has concluded various international climate pacts with heads of state and mayors from around the world in an effort to fight back against the federal government’s withdrawal from the Paris Agreement (Davenport and Nagourney 2017; Green and Jackel 2017). Countries like Canada and Mexico have already started acting on the policies prescribed by these pacts with California, in an effort to establish a carbon-cutting program, and California has approached China in hopes of a similar agreement. Also, California’s governor has attended the climate change meetings at the United Nations, participating in efforts to implement the Paris Agreement “with or without the US.”Footnote 44

The obvious next step would be to allow subnational units to formally join international treaties and conventions (albeit perhaps not on the same par with States). Where such subunits have the capacity to act autonomously and effectively within their own constitutional system, such a formal recognition would merely acknowledge a fait accompli: these units’ past and present efforts of promoting international law.

Some might object to this proposal on the ground that, by definition, public international law is international law by and for States and States only. But that is so purely as a matter of convention. There is no principled reason, internal to international law, why subnational units should not be takers of such law. For an example of how superordinate and subordinate units can both be parties to binding international agreements, consider the case of the European Union: both the EU and its member States can enter into treaties with other States. Public international law does not explain why subunits should generally be prevented from “operating as independent actors in the international legal system,” “nor does it acknowledge that the international legal system might have an interest in reforming the legal status of cities” or other subnational units (Frug and Barron 2006, p. 15).

The most straightforward way for subnational units to become makers and takers of international law would be to formally recognize the agreements that they conclude as public international law. As we saw, both states and cities have lately been very active in concluding voluntary global agreements across international borders. But at present those “global agreements” are sharply distinguished from international treaties; they do not give rise to any formal international obligations.Footnote 45 Were subunits to be acknowledged as makers and takers of international law, such agreements would acquire a different international legal standing, as “hard law” rather than the “soft law” that they presently constitute (Abbott and Snidal 2000).

Would such formal recognition make a huge difference?Footnote 46 International treaties are often not particularly “hard” law either, lacking any strong enforcement mechanisms; and much of the most valuable international cooperation and coordination comes through “soft law” mechanisms anyway (Swiney 2020). Furthermore, many of the soft-law agreements already being concluded among subnational units are substantively fairly impressive (at least on paper). Still, insofar as those agreements are purely “declaratory,” they can easily amount to “cheap talk.”Footnote 47 Subnational units being able to enter into agreements that are binding under international law might help to separate the wheat from the chaff in this respect, helping identify which subnational units are in earnest and which are not. This would make international cooperation and coordination easier.

To be sure, the question of formally acknowledging subnational units in international law deserves a separate discussion. Here, I merely wanted to draw attention to its sheer possibility. As we saw, subnational units already enjoy extensive autonomy and have legal instruments to pursue partially independent agendas on the international stage. A formally recognized role and responsibility for subnational units in the area of immigration and refugee protection may be hard to dismiss out of hand. If further empowering subnational units can bridge gaps in international law enforcement (as we have reason to believe), then subnational units should be allowed to participate more in international governance—if not on a par with nation States, then as a trusty second-best.