Attribution, Jurisdiction, Discrimination, Decapitation: A Comment on Makuchyan and Minasyan v. Azerbaijan and Hungary

You know how, every once in a while, you read a case that has everything? I mean really everything? Great facts. Grisly facts even, for those so inclined – say involving a beheading by a state agent. Great law. Not just some genuine legal innovation worthy of scholarly commentary – that’s fine obviously, but not all that uncommon. I mean proper, nerdy, esoteric legal stuff. It doesn’t have to be hugely important; it doesn’t have to concern the world’s most powerful states and its most pressing issues; but it still sticks in your head, it really does. And thank God it’s not about Covid. That’s the kind of case I mean.

So let me tell you about one such case that has everything – Makuchyan and Minasyan v. Azerbaijan and Hungary, a Chamber judgment of the European Court of Human Rights, decided last month. As we will see, while focused primarily on the substantive and procedural obligations of states arising from the right to life, the case also raises significant systemic questions of the European Convention’s extraterritorial application, in particular to assassinations by state agents of individuals abroad;  the proof of discriminatory intent in the right to life context; the attribution of the conduct of state organs not committed in official capacity; and, most interestingly, it also provides us with an exceptionally rare example of the application of the acknowledgment and adoption attribution standard in Article 11 of the ILC Articles on State Responsibility.

But first, the facts. In early 2004, two Armenian army officers and one Azerbaijani officer found themselves in Budapest, attending a NATO-sponsored English language course. One month into the course, R.S., the Azerbaijani officer (whom the two Armenians had allegedly mocked and otherwise provoked), went out and bought himself an axe and a sharpening stone. In the early morning hours of 19 February 2004, the anniversary of the start of the Nagorno-Karabakh conflict, R.S. killed one of the Armenian soldiers while he was asleep, decapitating him with at least twelve blows of said axe. He tried doing the same to the other Armenian officer, but the second victim was saved by a sturdy door and by the arrival of the Hungarian police, who apprehended R.S.

R.S. was then tried in Hungary and convicted for murder and attempted murder; Hungarian courts sentenced him to life imprisonment. In August 2012 R.S. was transferred by Hungarian authorities to Azerbaijan, after the Azerbaijani government had provided assurances that R.S. would serve the remainder of his sentence there. Immediately upon his arrival, R.S. was pardoned by the Azerbaijani president and given a hero’s welcome. He was also promoted in military rank, paid salary arrears and offered a state apartment. A special section of the official webpage of the Azerbaijani president was set up on which individuals (including numerous state officials) could express their congratulations on R.S.’ release and pardon.

The applicants before the European Court were the surviving Armenian soldier and a relative of the slain one. They complained that Azerbaijan violated Article 2 ECHR in both its substantive and its procedural aspects – the former because R.S., as a soldier in the Azerbaijani military and thus a state agent, killed one of his victims and attempted to kill the other; the latter because of his release from prison. The applicants additionally claimed a violation of the prohibition of discrimination in Article 14 ECHR, when read together with the right to life, because R.S.’ violence and the Azerbaijani authorities’ decision to release him were motivated by a discriminatory animus against Armenians. Finally, the applicants complained that Hungary had also violated the procedural limb of Article 2 by transferring R.S. to Azerbaijan.

In its judgment, the Court found a violation of Article 2 in its procedural aspect, as well as a violation of Article 14 by Azerbaijan. But it did not find a violation of Article 2 in its substantive aspect, on the grounds that the conduct of R.S. – i.e. the killing and attempted killing of the two Armenian officers in Budapest – was not attributable to Azerbaijan. Nor did the Court find a procedural violation of Article 2 by Hungary, holding that Hungary was entitled to rely on the assurances provided by the Azerbaijani government.

A procedural (and discriminatory) violation of the right to life

The Court’s finding of a violation of the procedural aspect of the right to life is obviously the most important feature of the case, its bottom line. There have been many other judicial pronouncements in Europe and elsewhere about the compatibility of amnesties and pardons with human rights. What’s noteworthy here is that the Court basically conducted a rigorous judicial review of an individualized pardon decision in its context, i.e. coupled with all of the affirmative acts of the Azerbaijani government, effectively finding that (para 172) all of these acts cumulatively granted R.S. impunity for his crimes (and were intended to do so). For a more extensive discussion of this aspect of the case, see this post by Cedric Ryngaert and Kushtrim Istrefi over on Strasbourg Observers.

Equally noteworthy is how the Court dealt with the non-discrimination aspect of the case, by finding a violation of Article 14 ECHR (read in conjunction with Article 2) on the basis that the decision of the Azerbaijani authorities to release R.S. was racially motivated by an animus against ethnic Armenians. It is rare for international human rights bodies to find allegations by claimants of a racially discriminatory animus to have been sufficiently substantiated (for one such important case where the animus was ultimately not proven see Nachova v. Bulgaria). But here the Court had no such difficulty. Relying on all of the approbatory conduct of the Azerbaijani government, including several official statements glorifying R.S.’ crimes and the special webpage set up for that purpose, the Court was (para 218)

satisfied that the applicants have put forward sufficiently strong, clear and concordant inferences as to make a convincing prima facie case that the measures taken by the Azerbaijani authorities in respect of R.S. were racially motivated. The Court is mindful of the difficulty faced by the applicants in proving such bias beyond a reasonable doubt, given that the facts in issue lie wholly, or in large part, within the exclusive knowledge of the Azerbaijani authorities. The Court considers that, given the particular circumstances of the present case, it was therefore incumbent on the respondent Government to disprove the arguable discrimination allegation made by the applicants.

This the Azerbaijani government then failed to do (paras 219-221). The Court had thus shifted the burden of proof of (lack of) discriminatory intent on the respondent government, because the applicants had established a ‘convincing prima facie case’ or an ‘arguable discrimination allegation.’ (In Nachova, by contrast, there was disagreement between the Chamber and the Grand Chamber as to whether it was appropriate to shift the burden of proof with regard to the violation of the substantive limb of Article 2; the Grand Chamber in the end declined to do so).

I will not, however, discuss these matters further, important as they are. Rather, I will turn to the two issues that really make this an extraordinary case: the question of the Convention’s extraterritorial application, and the issue of the attribution of R.S.’ conduct to Azerbaijan under the law of state responsibility.

Extraterritoriality

Recall that the applicants never set foot on Azerbaijani territory. The claimed substantive violation of the right to life – R.S.’ murderous rampage – happened in Budapest. The claimed procedural violation did occur in Azerbaijan, in the sense that the conduct of Azerbaijani authorities that comprised the violation took place there, but the victims of the violation were not there at the relevant time. Both violations, in other words, were extraterritorial. (The same goes for the alleged procedural violation by Hungary, the transfer of R.S., which again took place when the victims were no longer in Hungary).

Neither the applicants nor the two respondent governments raised the extraterritorial application issue, i.e. whether the applicants were within Azerbaijan’s jurisdiction in the sense of Article 1 ECHR. This is actually not at all uncommon; there are many examples of cases in which an extraterritoriality issue is reasonably explicit but the parties never raise it, probably because it would seem entirely arbitrary (as here) for the Convention not to apply. In most such cases the Court just ignores the whole extraterritoriality problem altogether – it’s somehow ‘obvious’ that the Convention applies. But not so here – the Court raised the issue proprio motu, but, curiously, did so only with regard to the obligations of Azerbaijan and not those of Hungary (paras 47-52).

Regarding the procedural limb of Article 2, for which it eventually did find a violation on the part of Azerbaijan, the Court invoked its recent Güzelyurtlu judgment (see also here at 48), to basically say that because R.S. was present on Azerbaijani territory and because Azerbaijan had sought his transfer it had a procedural duty under the ECHR to enforce his sentence. These were then ‘special features’ that triggered the applicability of the Convention. But this approach makes little sense if the jurisdictional inquiry is, as traditionally understood, one into the relationship between the state and the victim of the human rights violation. Like in Güzelyurtlu, this reasoning doesn’t really amount to much – the Court (rightly) feels that it would be arbitrary for the Convention not to apply, and is then compelled to come up with some ‘special features’ that render the Convention applicable now, when it wants the Convention to apply, while potentially imposing some limits in other, not exactly equivalent cases. For example, would the outcome of the jurisdictional inquiry have been the same if Azerbaijan had not itself sought R.S.’ transfer, or if R.S. was a fugitive on the run who somehow entered Azerbaijani territory – I have no idea, and I suspect the Court doesn’t either.

But the Court’s reasoning regarding the claimed substantive violation of the Convention, i.e. the killing and attempted killing by R.S. of the two Armenian officers, is even more interesting (if also laconic and evasive) because of its possible implications (para 52):

As regards the complaint under the substantive limb of Article 2, as formulated by the applicants in the present case, the Court considers that the issue of jurisdiction – that is to say, whether the victims were under the control of R.S. and whether he acted as an Azerbaijani State agent at the time of the crimes – is interlinked with the substance of the applicants’ allegations and shall be examined simultaneously with their complaint.

The issue here is basically whether the Convention would apply if an axe-wielding state agent chopped somebody to bits outside the state’s territory. Note how the Court has framed that issue through the personal conception or model of state jurisdiction as authority or control over the victim exercised by the state agent. This question of whether the Convention applies to kinetic uses of force abroad has been the most controversial and complex one in the Court’s extraterritoriality jurisprudence – from Bankovic (where the Court had ruled that dropping a bomb from the air on an individual did not suffice to create a jurisdictional link) to Al-Skeini (where the Court on the one hand accepted that killing an individual was an exercise of power or control over them, but on the other incorporated a ‘public powers’ limiting principle into its analysis, modifying Bankovic without overruling it entirely). Makuchyan and Minasyan is the first case in which the Court had to consider whether the killing or assassination of a single individual by a stage agent, outside any armed conflict, would be covered by the Convention.

I have long argued that the answer to this question must be yes, and that moreover negative obligations of restraint under human rights treaties should apply without any territorial restriction. There is simply no non-arbitrary way of limiting the personal conception of jurisdiction, which must include any kind of state conduct that deprives the victim of life, as English courts have realized in the Al-Saadoon cases. Whether it’s the 2006 assassination of Alexander Litvinenko by Russian agents in London using a polonium-laced teapot; the 2017 murder of Kim Jong-nam, the unfortunate half-brother of the North Korean dictator, by VX nerve argent smeared on his skin at the Kuala Lumpur airport; the 2018 Salisbury attack on the Skripals, resulting in the death of a bystander; the murder of Jamal Khashoggi; the 2019 killing of a Chechen dissident in a Berlin park; or indeed cyber attacks against hospitals during the pandemic – in all of these instances the obligation to refrain from using lethal force simply must apply.

But the Court is still afraid of saying so directly, because if it does it would become the final arbiter of any lethal use of force by a European state acting overseas (e.g. in Afghanistan). Now, however, the Court in Makuchyan and Minasyan left open the door that it might realign its jurisprudence with those of other human rights bodies, such as the Human Rights Committee in its General Comment 36. The Court did not say that the Convention would not apply, per Bankovic, to the killing of a single individual by a state agent – it could have dismissed the case in an Article 1 jurisdiction analysis, but it chose not to do so. Rather, assumed that the Convention could have applied.

On the facts of our case, however, it didn’t – because, according to the Court, the conduct of R.S. was at the time of the killing not attributable to Azerbaijan. Without attribution, there is no internationally wrongful act of the state (i.e. a violation of the right to life). But there is also no jurisdictional link, because the jurisdiction-establishing conduct (authority, power, control over the victim) and the violation-establishing-conduct (the axe) were one and the same. Thus, at the closing of its attribution analysis later in the judgment, the Court says that in the absence of attribution there was no violation of Article 2 ‘even assuming that Azerbaijan might be considered as having jurisdiction over R.S.’s actions in the particular circumstances of the present case’ (para 120, emphasis added – note however the confusion in framing jurisdiction as relating to the actions of R.S. as opposed to his control over the victims).

Attribution: conduct of state organs not acting in their official capacity

Why wasn’t then the conduct of R.S., an officer in the Azerbaijani military and thus a de jure state organ, not attributable to Azerbaijan? The Court here made two moves. First, it held that R.S., while a state organ, was not acting in his official capacity (para 111):

Turning to the present case, the Court attaches crucial importance to the fact that R.S., although a member of the Azerbaijani military forces at the material time, was not acting in the exercise of his official duties when he killed the second applicant’s relative and was preparing to kill the first applicant. In particular, he was not engaged in any planned operation or in a spontaneous chase (contrast Leonidis v. Greece, no. 43326/05, § 58, 8 January 2009). On the contrary, according to the record of the circumstances of the case established by the Hungarian courts, the crimes were committed as a result of R.S.’s private decision to kill during the night and outside of training hours the Armenian participants in the NATO-sponsored language course that they were attending because they had allegedly previously mocked and provoked him. It has not been suggested that the crimes committed by R.S. were committed on orders given by his superiors and nor is there is any evidentiary basis for such a far-reaching conclusion.

This is the right result, but the Court’s reasoning is somewhat problematic. On this point the Court does not engage at all with the ILC Articles on State Responsibility, in particular with the commentary to Article 4 ASR (at 42, para 13, on conduct committed in official as opposed to private capacity) and Article 7 ASR (on the attribution of ultra vires acts, when these are committed in official capacity). The only authority it relies on (Leonidis) is its own and is entirely irrelevant, and when it does so its reasoning is contradictory (in effectively saying that R.S. was engaging neither in planned nor in spontaneous activity). Finally, the fact that R.S. was not acting under orders from his superiors (i.e. that he acted ultra vires) does not necessarily imply that his act was private – for example, a police officer who tortures a detainee contrary to domestic law and superior orders is nonetheless acting in his official capacity.

That said, again the Court’s conclusion on this point is basically sound – R.S. was attending a language course in a foreign country and was not performing any other official function on behalf of Azerbaijan at the relevant time (e.g. he was not a security guard at the embassy killing someone on duty). I do not therefore wish to nit-pick any further regarding the Court’s reasoning above, save to say that there are numerous other cases in which the Court fails to appreciate the difference between the private conduct of a state organ and ultra vires official conduct, or fails to engage with the ILC Articles – for an extended discussion see this piece on special rules of attribution in international law that I recently posted on SSRN, esp at 74-75.

Attribution: conduct acknowledged and adopted by the state as its own

This brings me to the second attribution rule applied by the Court, on which it actually extensively engaged with the work of the ILC, probably more so than in any of its other judgments: that a state will be responsible even for conduct that was initially purely private if it subsequently acknowledges and adopts that conduct as its own, per Article 11 ASR. The first key feature of the attribution rule in Article 11 ASR is its retrospectivity – at the time the wrongful conduct was committed it was not attributable to the state, but it later becomes attributable due to the state’s own decision to adopt the conduct as its own. The second key feature is that Article 11 requires more than mere state endorsement of some wrongful conduct – the state must affirmatively make that conduct its own (ASR commentary, at 53, para 6).

The classical case on this rule is the ICJ’s judgment in Tehran Hostages. There the ICJ considered that the hostage-taking in the US embassy in Tehran was initially not attributable to Iran (which was responsible for its own omission, its failure to protect the embassy) but then become attributable to Iran, primarily on the basis of several statements by Ayatollah Khomeini, including a formal decree setting out the state’s policy on the matter (paras 71-74).

The rule in Article 11 ASR was meant to apply only exceptionally. It is therefore no wonder that only a very few cases have dealt with it after the adoption of the Articles. One, which the European Court itself cites, is a strange tangent in an ICTY decision in which the Tribunal used Article 11 by analogy to determine whether the apprehension of a suspect was attributable to NATO peacekeepers in Bosnia, while hedging its discussion with caveats regarding the customary status of the rule and the authority of the ILC Articles as a whole.

There have however been at least five investment arbitral awards – not cited by the European Court – in which the tribunals fully accepted the customary status of Article 11 and its requirements. In none of these cases was Article 11 the linchpin of the attribution analysis; attribution was generally already made out by the tribunal on the basis of Articles 4 or 5 ASR, only venturing into Article 11 as a backup, alternative argument. Thus, in Luigiterzo Bosca v. Lithuania (2013), Article 11 features only in a footnote (para 128 n 114). In Bernhard von Pezold and Others v. Republic of Zimbabwe (2015) Article 11 got slightly more detailed treatment, with the tribunal concluding that mere endorsement was insufficient for the application of the rule (paras 447-449); the same thing happened in Unión Fenosa Gas, S.A. v. Arab Republic of Egypt (2018) (paras 9.120-121). Two decisions contain more detailed factual analysis, again while entirely embracing the ILC’s approach –  Clayton and Bilcon of Delaware Inc. v. Government of Canada (2015), paras 321-324; Saint-Gobain Performance Plastics Europe v. Bolivarian Republic of Venezuela (2016), paras 461-466.

The latter two cases, in which the tribunals found that the requirements of Article 11 were satisfied, followed the Tehran Hostages model: there was a formal decision of some kind by a governmental authority that adopted the relevant conduct as the state’s own. And that’s what was effectively missing in Makuchyan and Minasyan before the European Court – yes, R.S. got a presidential pardon, a hero’s welcome, a promotion, salary arrears, a flat. Clearly the Azerbaijani state approved or endorsed what he had done (paras 114-117). But Article 11 requires more (para 118):

Having most thoroughly examined the nature and scope of the impugned measures within the overall context in which they were taken and in the light of international law, the Court is unable to conclusively find that such “clear and unequivocal” “acknowledgement” and “adoption” indeed took place. In substance those measures can be interpreted not so much as the State’s “acknowledgment” and “adoption” of R.S.’s criminal acts, as such, but rather as having the purpose of publicly addressing, recognising and remedying R.S.’s adverse personal, professional and financial situation, which the authorities of Azerbaijan perceived, unjustifiably in the Court’s view, as being the consequence of the allegedly flawed criminal proceedings in Hungary (see paragraph 106 above). It follows that, although the Court considers it beyond any doubt that by their actions various institutions and highest officials of the State of Azerbaijan “approved” and “endorsed” the criminal acts of R.S., applying the very high threshold set by Article 11 of the Draft Articles – as interpreted and applied by international tribunals, in particular the ICJ and the ICTY (see paragraphs 36-37 above) – the Court cannot but conclude that, on the facts of the case, as presented by the applicants, it has not been convincingly demonstrated that the State of Azerbaijan “clearly and unequivocally” “acknowledged” and “adopted” “as its own” R.S.’s deplorable acts, thus assuming, as such, responsibility for his actual killing of G.M. and the preparations for the murder of the first applicant. The Court places emphasis on the fact that this assessment is undertaken on the basis of the very stringent standards set out by the existing rules of international law, as they stood at the material time and stand today, from which the Court sees no reason or possibility to depart in the present case. Therefore, contrary to what was argued by the applicants, the present case cannot be considered fully comparable to the ICJ’s United States Diplomatic and Consular Staff in Tehran case.

To my mind, the Court’s analysis is in line with the approaches of the ILC and the ICJ. Of course, one could argue that the ILC’s standard is too demanding, as Cedric and Kushtrim do in their post. However, not even the dissenting Judge Pinto de Albuquerque thought so; his disagreement with the Court was on the application of the law to the facts, rather than on what the requirements of the attribution rule were. But in my view one can still draw a reasonably clear line between an expression of state approval for a certain course of conduct and an expression of the state’s will to adopt that conduct as its own – a line that this case well illustrates.

Perhaps Makuchyan and Minasyan will go to the Grand Chamber, as Judge Pinto suggests it should. But I doubt that the Grand Chamber’s judgment will be very different. In particular it is normatively difficult to justify why a simple expression of approval or endorsement by state officials should result in the retrospective attribution of the approved conduct – one can approve of many things without necessarily assuming responsibility for doing those things. In any event, there is no doubt that Makuchyan and Minasyan will find its way into international law textbooks, as the case containing the most detailed analysis of Article 11 ASR to date, and as an excellent comparator to Tehran Hostages.

Didn’t I say this was a great case?