The Absence of Practice Supporting the “Unwilling or Unable” Test

by Kevin Jon Heller

Regular readers of the blog know that one of my hobbyhorses is the “unwilling or unable” test for self-defense against non-state actors. As I have often pointed out, scholars seem much more enamored with the test than states. The newest (regrettable) case in point: my friend Claus Kress, who is one of the world’s best international-law scholars. Here is what he writes in an otherwise-excellent contribution to Just Security about the use of force against ISIL in Syria (emphasis mine):

It therefore follows not only from the right of self-defense’s general requirement of necessity, but primarily from the respect for the sovereignty of the territorial State that the right of self-defense in case of a non-State armed attack is of a subsidiary nature. It presupposes that the territorial State is either unwilling or unable to end the non-State armed attack – or, as it should be added for the sake of completeness, fails to exercise due diligence to that effect. State practice is remarkably consistent with these principles. As Professor Ashley Deeks has demonstrated in a formidable article, the legal claims to a right of collective self-defense in cases of non-State armed attacks have generally included the statement that the territorial State is unwilling or unable to deal with the non-State threat.

In terms of what the “unable or unwilling” test might look like if it represented customary international law, Deeks’s article is indeed excellent. But the article is anything but “formidable” in terms of state practice that supports the test. Indeed, the non-state actor section of the article spans all of two pages (pp. 501-03) — and cites precisely two states that officially endorse “unwilling or unable”: the United Kingdom and the United States. That’s it. And those are the same two states that Claus discusses in his post.

Simply put, there is simply no “consistent practice” that supports the “unwilling or unable” test, and scholars need to be careful not to put states in the “unwilling or unable” camp simply because they are willing to use armed force against a non-state actor. Deeks has been particularly prone to this kind of overinclusiveness, most recently arguing that Jordan, Bahrain, Qatar, the UAE, and Iraq support the “unwilling or unable” test because they have attacked ISIL in Syria — this despite the fact that all five states are members of the Arab League, which has specifically rejected the test in the context of Israel’s attacks on Hezbollah in Lebanon. (Actual opinio juris.)

I have the utmost respect for Claus, and I have no desire to pick on Deeks. But methodological rigor is particularly critical when it comes to doctrines like “unwilling or unable,” because its actual adoption by states would open the floodgates to the extraterritorial (ie, sovereignty-infringing) use of force against non-state actors. There may well come a time when the “unwilling or unable” test reflects customary international law, but that time is not now. Two states do not a customary rule make, however powerful those states may be. And we cannot simply ignore the states in the Global South, however inconvenient powerful states in the Global North may find their views.

13 Responses

  1. Yes, “unwilling or unable” is not “the” “test.” Article 51 self-defense against non-state actor armed attacks is permissible when the armed attacks are underway — whether or not the state from which they emanate (if they emanate from the territory of a state) is unwilling or unable. While the armed attack is ongoing, the foreign state is, nonetheless, per se unable. http://ssrn.comabstract=1701992 — and see the extensive list of textwriters, etc. in footnote 3.

  2. Kevin: Concededly, deciding *when and how* a host state is unwilling or unable to address the threat raises a host of thorny issues — the subject of the bulk of Ashley’s fine article. But I don’t quite understand why you resist the notion that there is an U/U test at all.

    Doesn’t it follow quite naturally from the basic ad bellum requirements for self-defense?

    The basic rule under 2(4) is that the threatened state must refrain from using force “against the territorial integrity” of the host state. There’s an exception for self-defense, traditionally defined–a standard that requires, inter alia, that the breach of territorial integrity be necessary and proportionate to the need to prevent further attacks from the host state itself or, as here, from a nonstate group taking sanctuary in that state. If the host state is willing and able to deal with the problem, then the use of force is, virtually by definition, not “necessary.” But if the host state is unwilling or unable, then the use of force is-or at least might be–necessary to stop the threat. (Of course, there might be other reasons why it is not necessary or proportionate, as well–e.g., the attacking group no longer has attack capabilities. The fact that the host state is unable or unwilling is a necessary, but not always a sufficient, predicate for the self-defense exception.)

    What am I overlooking in this simple account? And which states do *not* agree with this basic idea — and would conclude that the threatened state must instead stand by and absorb the attacks from the host state? Seems unlikely, especially in light of the history of analogous situations. Indeed, U/U (more than “imminence”) was at the heart of the actual dispute between the U.S. and England in the Caroline incident: Both parties agreed that England could act if the U.S. could not or would not stop the rebel attacks from its territory; what divided the parties was whether U.S. action against the rebels was actually forthcoming and sufficient, i.e., whether the U.S. authorities actually were unwilling or unable.

  3. Thank you for this piece. I have couple of questions on methodology. First, is the search for opinio juris actually what is relevant here? It would seem more appropriate to look simply for state practice as a supplementary means of interpretation under article 32 to the VCLT (article 31(3)(b) being irrelevant as it would necessitate practice of UN members) as a means of interpreting the UN Charter (see also page 59 in this piece: In this regard, the current practice of the Arab League states is very pertinent.
    Second, indeed the Arab League condemned what it termed Israeli “aggression” against Lebanon. However, from the statements of the Arab League and its member states in the UNSC debates (cited by Ruys), it is actually unclear why they term it aggression. It is very plausible that the reason for the use of the word aggression actually relates to perceived disproportionate use of force (and not the fact that forced is being used following an attack by a non-state actor).

  4. Would it not appear that practice relevant for interpretation requires it to be undertaken “in the understanding of the parties as to its interpretation”? In other words, mere practice is not sufficient, there needs to be a subjective conviction that such particular practice constitutes what the rule (as interpreted) calls for. See also ILC, Second Report by Special Rapporteur Nolte, Draft Conclusion 6, A/CN.4/671.

    On a different note, the importance of the requirement of necessity has already been mentioned. Much of the debate on U&U could be seen through the lens of necessity: is it necessary – as in, last resort, no alternatives – to act in self-defence against the territorial State when it is unwilling or unable (see also Gray’s chapter in Evans’ International Law)? Yet, a distinction between inability and unwillingness is often not made. Unwillingness is one thing, but inability calls for the victim State offering assistance to (and thus requesting consent from) the territorial State. Only when this is rejected the territorial State becomes unwilling. In other words, inability by and of itself should not be seen on par with unwillingness.

  5. Marty,

    Your comment assumes precisely what is in dispute: that self-defence is available against an NSA whose armed attacks are not attributable to the territorial state. It may seem self-evident that states should not have to “absorb the attacks” from the NSA, but that is a very post-9/11 view. Prior to 9/11, there seems to be little question that self-defence did not permit violating the sovereignty of a state — even an “unwilling or unable” one — that was in no way responsible for the armed attacks by an NSA using its territory as a base. Here, for example, is how Tams, a leading scholar in the area, describes the situation:

    Self-defence was the principal ground on which states relied in order to justify their use of anti-terrorist force. The underlying argument was straightforward: setting out a broad construction of self-defence, states claimed a right to respond to attacks even if these were not carried out by another state. While these claims were made frequently, they were never received favourably by the international community. In fact, during the 1970s and 1980s, the international community rejected them almost systematically. To give but some examples, Israel’s 1985 raid on the PLO Headquarters out- side Tunis was ‘condemn[ed] vigorously’ by the Security Council, which declared it an ‘act of armed aggression … in flagrant violation of the Charter of the United Nations’ and urged other states ‘to take measures to dissuade Israel from resorting to such acts against the sovereignty and territorial integrity of all States’. Similarly, the United States’ 1986 raid on targets in Libya, while controversially discussed by commentators, was roundly rejected by the General Assembly as ‘a violation of the Charter of the United Nations and of international law’. South Africa’s incursions into neighbouring states, if anything, met with stiffer resistance.


    First, self-defence against armed attacks by non-state actors was admitted in principle, but only under narrow conditions. For an attack to qualify as an ‘armed attack’ in the sense of Article 51 (or its customary equivalent), the direct attack by a non-state actor had to be attributed to another state under rather stringent rules on attribution. The law on this point was shaped by the ICJ’s judgment in the Nicaragua case, which concerned the relationship between a state and rebel forces, but came to define the rules on attribution generally. In that decision, the Court (drawing on the General Assembly’s Definition of Aggression) accepted that the jus ad bellum could be violated by ‘the sending by or on behalf of a state of armed bands, groups, irregulars or mercenaries, which carry out acts of armed force against another state’. Yet for the conduct of irregular forces to be attributable to a state, that state had to exercise ‘effective control [over] the military or paramilitary operations’ in question, whereas logistical or other support was insufficient. Self-defence thus depended on complex, and typically fact-dependent, questions of attribution, and required responding states to show a substantial involvement of the territorial state in the very attacks of a terrorist organization against which the response was directed (referred to as ‘effective control’ test). As a consequence, only terrorist attacks effectively controlled by another state triggered a right of self-defence. By adopting a restrictive approach to attribution the Court effectively restricted self- defence to the inter-state context. This approach seemed in line with an inter-state reading of the jus ad bellum, took into account the scepticism among UN members against broader readings of self-defence (which would have allowed the abuse of the concept), and for a while was hardly attacked as a matter of principle.

    The customary “baseline” post-9/11, then, is that self-defence against an NSA still requires the NSA’s attacks to be attributable to the territorial state — attribution requiring more than simply unwillingness or inability. To be sure, as with all custom, that rule can change, and could have changed, over time. I am certainly not saying that “unwilling or unable” could never be the customary rule. But the onus is on those who defend “unwilling or unable” to show sufficient state practice and opinio juris to establish that attribution is no longer required. Deeks has not done so — and I don’t think could do so. Tams, for example, says that although the ICJ’s “effective control” standard in Nicaragua no longer reflects custom, some kind of attribution is still required — something akin to aiding and abetting, in his view. But that is still not the “unwilling or unable” standard.

  6. Remy, the concern of Special Rapporteur Nolte in draft conclusion 6 seems to be focused on what he termed “voluntary practice” beyond what is demanded by the treaty (page 6 to that report). Conversely, when practice would otherwise constitute a violation of the agreement, it seems to be possible to presume the party considers its practice in accordance with that agreement. For instance, this appears to be the position of the ICJ in the Nuclear Weapons advisory opinion (para. 55).

  7. Marty: Article 51 is not limited to a violation of Article 2(4) in any way (e.g., there is no requirement that a member state engage in the armed attack or have imputed responsibility for such). Additionally, there is a requirement in Article 51 that an armed attack be underway (if an armed attack occurs). See Therefore, a mere threat of attack is not the trigger. Once the armed attack is underway, a state from which the armed attack emanates is necessarily “unable” to prevent the attack that is underway. Necessity for a response has been met. With due respect, the nature of the dispute between the UK and the US in the Caroline Incident is set forth in the above click-on. Yes, both were in agreement that once armed attacks by a non-state actor were occurring, there was a right to use measures of self-defense against the non-state actors located within the U.S., but the U.S. pushed for a necessity limitation of the means of response (esp. since the UK could have waited in those days to seize the vessel when it ventured into Canadian waters, etc.). Note the attention to a non-state actor battery firing across the border.

  8. Tams is absolutely incorrect re: a supposed “effective control” test for attribution, as the click-on in the above reply demonstrates. The ICJ used that test regarding responsibility for law of war violations. The ICJ recognized a “substantial involvement” test for imputation. Also, re: Israel example, Israel had claimed a right of “reprisal,” which the community does not accept. The U.S. 30 min. war against Libya was based on a claim of self-defense against Libya (not a NSA) under a calim of attribution.
    Re: the Caroline incident and several others in the 19th Cent., there was no need for attribution — none at all! The UK and US addressed the right of self-defense by an NSA that had been engaged in ongoind armed attacks.

  9. Jordan,

    You might try reading what Tams actually wrote: “The law on this point was shaped by the ICJ’s judgment in the Nicaragua case, which concerned the relationship between a state and rebel forces, but came to define the rules on attribution generally.” That is absolutely correct.

  10. what I addressed was the statement that “that state had to exercise ‘effective control…,'” “the ‘effective control’ test,” “only terrorist attacks effectively controlled by another state.” This is the wrong test.

  11. Dear all
    interesting debates – reflecting once more how the scope of self-defence continues to be contested. The article from which Kevin quotes above (Tams, Use of Force against Terrorists, EJIL 2009) seeks to provide historical context – which in my view remains essential as we appreciate the current challenges. For those who want to read further, the article is freely available, thanks to the generous EJIL editorial policy — have a look

    Perhaps I can add that the specific issue debated between Kevin and Jordan (and on which Kevin disagreed with Claus Kress in the initial post) — viz. should self-defence be available against non-State attacks as such? — gave rise to a follow-up between Kimberley Trapp and myself: both pieces are only a few pages long, but if read together, I think they give a useful general picture of the debate: see (Kimberley) and (me). With the distance of a few years time, I think the main point to take away from the debate about anti-terrorist self-defence is that no side should claim too much conceptual clarity: we all seem to be muddling through somehow.

Trackbacks and Pingbacks

  1. Boca Raton CPAt test review

    Opinio Juris » Blog Archive The Absence of Practice Supporting the “Unwilling or Unable” Test – Opinio Juris

  2. Lynell

    Opinio Juris » Blog Archive The Absence of Practice Supporting the “Unwilling or Unable” Test – Opinio Juris