By Tobias Thienel

Some time ago, I reported here the case of Saadi v. Italy, decided by the Grand Chamber of the European Court of Human Rights (ECtHR) on 28 February. That case is most interesting for its clear – and, I would say, entirely correct, welcome and timely – clarification that the prohibition on torture and inhuman and degrading treatment is indeed absolute, even in the context of expulsions and extraditions. My earlier post was about that bit. But the case also contains some interesting observations on the law of evidence. I turn to those now, somewhat foreshadowing what I am going to have to say on this and other questions in a forthcoming article in the German Yearbook of International Law (on ‘The Burden and Standard of Proof in the European Court of Human Rights’). As will be recalled, the case was all about the prohibition of removing someone from the territory if that person faces a risk of being subjected to inhuman treatment in the State to which he is sent (Soering v. United Kingdom, paras. 85-91; Chahal v. United Kingdom, paras. 79-82). Of course, the question then arises: how do you prove that there is such a risk? First, the burden of proof. The general principles are not in much doubt, although it depends what you mean by that ‘burden’. As has been rightly noted, ‘[t]he term "burden of proof" is one of the "slipperiest member[s] of the family of legal terms"’ (Schaffer ex rel. Schaffer v. Weast, 126 S.Ct. 528, 533-34 (2005) [link: slip op., p. 6]). In the ECtHR, the general position is this: if in the end, the Court is not persuaded that there is a risk of future ill-treatment abroad, the applicant will lose. The necessary case will not have been made. However, it is not exclusively for the applicant to present the evidence necessary to support his case. The Court will, instead, consider all the evidence before it, no matter what its source, and if necessary, the Court will make its own inquiries (Ireland v. United Kingdom, para. 160; Artico v. Italy, para. 30). There is, then, in common law terms, no ‘burden of producing evidence’, but there is a ‘burden of persuasion’. In principle, the latter is on the applicant, as concerns the question of the ‘real risk’. Whether a prima facie showing of a ‘real risk’ then shifts the burden to the respondent State, or is to be treated as strong evidence making the applicant’s case all but irresistible, with the tactical consequence that the State had better lead evidence in rebuttal, is an open question. Judge Zupan?i? in his Concurring Opinion in Saadi (para. 1) suggests the former should be done. On a similar question, where the ECtHR has long spoken of such a shift in the burden of proof to the State, the English Court of Appeal has read this as referring to a tactical burden (in the above sense), not to the ‘burden of persuasion’ as a matter of law: Sheppard v. Secretary of State for the Home Department [2002] EWCA Civ 1921, para. 13, commenting on Salman v. Turkey, para. 100. The difference between the two approaches is hardly very pronounced. But what about the standard of proof? Saadi has had a few things to say on that as well. The first error – for that, I think, is what it is – has not been made by the Court, but by counsel for the United Kingdom (intervening) at the oral hearing. As I mentioned in my first post on Saadi, the UK took issue with the threshold of the prohibition of expulsion. It said the test, as regards persons prejudicial to national security, should not be only that there is ‘real risk’ of ill-treatment. The test should be higher. Now, in making these arguments, counsel kept referring to the ‘real risk’ test as a ‘standard of proof’. Presumably against this background, he suggested the test should be that of the ‘balance of probabilities’ (which is, of course, the civil standard of proof at common law). With respect, the ‘real risk’ test is not too low a standard of proof. In fact, it is not a standard of proof at all. The contrary view has been taken (cf. I.N.S. v. Stevic, 467 U.S. 407, 421 (1984); I.N.S. v. Cardoza-Fonseca, 480 U.S. 421, 425 (1987)), but it confuses the finding of fact to be made (usually a matter of substantive law) with the certainty required to make that finding, as determined by the standard of proof (a procedural matter) (see Metropolitan Stevedore Co. v. Rambo, 521 U.S. 121, 137-8, n. 9 (1997)). The substantive law forces a court to decide if certain tests have been met on the facts, and the standard of proof tells the court when it may and must find such facts on the evidence. In the end, the court’s order (whether one or the other party may or may not do something, or owes the other something or not) follows from the rule of substantive law that the court has applied. Therefore, the ‘real risk’ test, satisfaction of which makes an expulsion unlawful, is a test of substantive law. So, the ‘real risk’ test is not a standard of proof. But is it possible to apply a standard of proof to the question of a risk, i.e. of the likelihood of future events? Judge Zupan?i? addresses the question in his Concurring Opinion in Saadi. The Judge argues that the paradigm of the legal process, and of the law of evidence that comes with it, is always retrospective: the courts are usually concerned with past events, and therefore with finding out whether past events have occurred or not. The ‘real risk’ test of Soering and Chahal, on the other hand, requires the Court to make a prospective assessment: ‘it applies to the probability of future events rather than something which has already happened’ (Judge Zupan?i?’s Concurring Opinion, para. 1). The Judge goes on to say: ‘It is therefore at least inconsistent to say that a certain standard of proof [?] could be applied. The simple reason for that is, of course, that one cannot prove a future event to any degree of probability because the law of evidence is a logical rather than a prophetic exercise.’ A similar point has been made by Lord Hoffmann in the House of Lords in Secretary of State for the Home Department v. Rehman [2001] UKHL 47, [2003] 1 AC 153, para. 56, where his Lordship says that ‘[i]n a criminal or civil trial in which the issue is whether a given event happened, it is sensible to say that one is sure that it did, or that one thinks it more likely than not that it did.’ However, the question in the case before the House in Rehman, whether the appellant posed a risk to national security, was one of ‘evaluation and judgment’. There is much to be said for this line of argument. A risk-based test requires evidence going to the probability of future events. A standard of proof, similarly, establishes a degree of likelihood that a court must find established before it can say that an event has occurred. It follows that evidence going against an applicant’s case that a ‘real risk’ exists might equally be described as suggesting a low risk (and not a ‘real risk’, in substantive law terms), or as suggesting a low likelihood of a ‘real risk’ (in terms of the applicable standard of proof). Defining a standard of proof therefore does not add much to the definition of a court’s exercise in establishing the existence or otherwise of a ‘real risk’. But the standard of proof is not completely without meaning even on this question. As the English Court of Appeal has said in considering Lord Hoffmann’s dictum in Rehman, a judge may make a finding of a risk with a greater or lesser degree of certainty. The standard of proof then comes in to tell the judge which degree of certainty is required to make the finding, and to base the resolution of the case on it (R. (N) v. Mental Health Review Tribunal (Northern Region) [2005] EWCA Civ 1605, [2006] QB 468, para. 99). That clarification may have some – albeit limited – usefulness in borderline cases. I therefore do not think the concept of a standard of proof is logically and necessarily misplaced in the assessment of a ‘real risk’. One question remains: what is the applicable standard of proof? In Soering, Chahal and many other cases, the general formulation of the prohibition of expulsion is formulated as applying ‘where substantial grounds have been shown for believing that the person concerned, if deported, faces a real risk of being subjected to treatment contrary to Article 3’ (formulation taken from Saadi, para. 125). I have already tried to show why I do not think the ‘real risk’ bit is a standard of proof. But the ‘substantial grounds for believing’ part has been taken as the applicable standard of proof. Indeed, the Court in Saadi itself took this view (paras. 140, 142). This stands in contrast to the Court’s view in Shamayev and 12 Others v. Georgia and Russia. There, at para. 353, the Court applied its general standard of proof (as applied in cases on the proof of past events) of ‘proof beyond a reasonable doubt’ also to the ‘real risk’ question. While I think the standard of ‘proof beyond a reasonable doubt’ is too demanding, and can hardly be defended (more on that in my article), I find it understandable that the same standard of proof should apply to the question of a ‘real risk’ as to any other question of fact. I cannot see why a separate and special standard of proof should be defined for the ‘real risk’ question, where the standard only has a very minor role to play anyway. The Court and Judge Zupan?i? in Saadi have aired some important evidentiary issues surrounding the Soering question of a ‘real risk’. The discussion is ongoing.