Autumn Ellis: Lawfulness of policies of public bodies and Freedom of Expression under Article 10 ECHR

The reaffirmation of the Gillick test by the Supreme Court

Thirty five years after Gillick v West Norfolk and Wisbech AHA (Gillick) was decided, the Supreme Court took the opportunity, in  R (A) v Secretary of State for the Home Department (A) and R (BF (Eritrea)) v Secretary of State for the Home Department(BF), (previously discussed in this blog here),  to restate the boundaries of the test for the lawfulness of policies published by public bodies, and to identify as erroneous cases which had relied on ‘other principles’ (A at [54]). Lords Sales and Burnett, giving the leading judgment in both cases, drew a distinction between policies which can be regarded as ‘sanctioning’ (by statement or omission), and those which are simply capable of ‘leading’ to, unlawful decision-making. They summarised the Gillick test as follows: ‘Does the policy in question authorise or approve unlawful conduct by those to whom it is directed?’ (A at [38]) (referred to here as the ‘authorisation/ approval test’). Distinct formulations of the lawfulness test relied on in previous cases, which turn on whether a given policy can be regarded as ‘leading’ to an ‘unacceptable risk’ of unlawful decision-making (referred to here as the ‘unacceptable risk test’), were incorrect to the extent that they constituted a departure from Gillick (A at [75]). 

In the recent Court of Appeal case of R (Harry Miller) v The College of Policing (Harry Miller), the appellant succeeded under the more onerous ‘authorisation/ approval’ test, where he had failed at first instance under the ‘unacceptable risk’ test. 

The application of the Gillick test in Harry Miller

The facts in Harry Miller may be briefly stated. Between November 2018 and January 2019, Mr Miller posted a series of tweets expressing ‘gender critical’ beliefs. Following a complaint by a transgender member of the public who had seen these tweets, Humberside Police: (i) made a record of a non-crime hate incident pursuant to the Hate Crime Operational Guidance (HCOG) issued in 2014 by the College of Policing (the CoP); and (ii) took various further actions (including visiting Miller at his place of work, and advising him that future tweets could lead to criminal prosecution). Mr Miller subsequently brought a claim challenging the lawfulness of: (i) the CoP’s HCOG; and (ii) Humberside Police’s further actions, relying on his right to freedom of expression under Article 10 of the European Convention of Human Rights (ECHR). At first instance, Mr Miller succeeded only on the latter claim. His appeal in relation to the unlawfulness of the HCOG succeeded in the Court of Appeal. 

Different tests for lawfulness of policies issued by public bodies

Mr Miller’s challenge to the lawfulness of the HCOG focused on its requirement of ‘perception based recording’. In particular, pursuant to Section 1.2, non-crime hate incidents were defined as those which were ‘perceived by the victim or any other person’ to be motivated by hostility or prejudice against a person who falls, or is perceived to fall, within one of the monitored hate crime strands (disability, race, religion, sexual orientation, and transgender). Pursuant to Section 6.3, a mandatory recording obligation arose following a report of such an incident ‘irrespective of whether there is any evidence to identify the hate element…’ Only limited exceptions were specified, and the policy contained no (express) discretion to exclude irrational reports. 

In the High Court, relying on the Court of Appeal judgment in BF, Mr Justice Knowles identified the question to be addressed on the lawfulness of the HCOG as: ‘[T]he Claimant must show that HCOG creates a real risk of more than a minimal number of cases where Article 10(1) will be unlawfully infringed’ (at [219]) . In the Court of Appeal Dame Victoria Sharp, with the benefit of the Supreme Court rulings in A and BF, applied a different test noting that: ‘assessing the lawfulness of a Policy by reference to whether it creates a real risk of unfairness in a more than a minimal number of cases, insofar as that test departs from the more rigorous test to be derived from Gillick…was incorrect and should not be followed’ (at [105, ftnt 7]). On this basis, the Court identified the correct question as: ‘Does the Guidance sanction or positively approve or encourage unlawful conduct viz, conduct which violates Article 10?’ (at [106]). 

Different outcomes on the application of the tests

Pursuant to the ‘unacceptable risk’ test, Knowles J had found the HCOG to be fully aligned with the right to freedom of expression as encapsulated in Article 10, and thus incapable of generating a risk of the relevant type.

In contrast, pursuant to the ‘authorisation/ approval’ test, the Court of Appeal found that the HCOG failed to meet the requisite standard. The Court’s analysis differed from that at first instance in the following respects. First, the ‘mere recording’ of a non-crime hate incident, without more, was held to be sufficient to constitute an interference engaging Article 10(1). In this regard, the Court noted that in the Article 10 context ‘special protection is afforded to political speech and debate on questions of public interest’ (at [69]), and the recording of Mr Miller’s tweets as a non-crime hate incident, and of their potential disclosure on an Enhanced Criminal Record Certificate, was capable of having a ‘chilling effect’ on future similar statements (at [76]). Secondly, although the Court agreed that this interference complied with the first two of the justificatory requirements of Article 10(2) (prescribed by law, and in pursuit of legitimate aims: the prevention of crime, and the protection of the rights of others), it did not consider that the third requirement of proportionality was met as assessed by reference to the four stage test set out in Bank Mellat v HM Treasury (No 2) (Bank Mellat) (at [74]).  In this respect, the HCOG met the first two limbs of the Bank Mellat test (its objectives were sufficiently important to justify the limitation of a protected right, and it was rationally connected to those objectives). However, the Court found that it did not meet the last two limbs, which substantially overlapped: the unavailability of less intrusive means, and an overall balance in favour of the HCOG’s objectives when weighed against the severity of its effects.The following key points were made in reaching the conclusion that the HCOG was disproportionate. 

1. A court must make its own assessment of the availability of less intrusive means: At first instance, Knowles J placed strong reliance on the fact that the HCOG had been derived from ‘sources which should command great respect and weight’ such as the Macpherson Report of February 1999. As such, the measures identified in the HCOG represent ‘what is thought necessary’ to achieve its aims (at [228]). The Court of Appeal regarded this approach as erroneous: notwithstanding the authority of HCOG’s sources, the court ‘is obliged to consider for itself whether any less intrusive measure could have been used’ (at [104]).
2. Exceptionally wide net for non-crime hate incidents: The Court took into account the different treatment under the HCOG in relation to non-crime hate incidents and hate crimes. With regard to the former, it noted the  ‘extraordinary’ breadth of the recording obligation, arising as a combination of the perception based recording trigger, a ‘low threshold’ for the definition of ‘hostility’, and the limited number of permissible exceptions. As a result of its broad scope, incidents could be captured that were both ‘non-crime’ and ‘non-hate’ in nature. In contrast, the sections of the HCOG relating to hate crime required an ‘objective assessment of the evidence’ in relation to both the ‘criminal’ and ‘hate’ elements of the relevant conduct (at [111 to 113]).
3. Limited capacity of police to investigate such incidents: The Court went on to acknowledge the ‘critical importance’ of perception-based recording, deriving from the Macpherson Report. However, it pointed out that that Report had been restricted to racial non-crime hate incidents, as opposed to all five of the now monitored strands; and it had also included the ‘equally strong recommendation that criminal and non-criminal hate incidents should be investigated with equal commitment’. Changes in how people communicate since the Macpherson Report, as a result of the internet and social media, means this equality in terms of investigation is unlikely to be achievable given the limited resources of the police (at [114]).
4. ‘Chilling effect’ on legitimate debate: The Court referred to the broader context of the case: ‘the public debate concerning transgender issues’. Echoing its earlier discussion on the issue of interference, it noted the implications of the HCOG for freedom of expression within this debate: ‘specifically, the chilling effect that it has on an important issue of legitimate public interest’. It further noted the potentially stigmatising effect of the language used in the recording obligation. In terms of safeguards in the legal framework relating to the retention, recording and disclosure of information by the police, the Court noted that these were primarily addressed to Article 8, rather than Article 10, rights; and that subsequent revisions to the HCOG do not go ‘nearly far enough to address the chilling effect of perception-based recording’ (at [115 to 119]).
5. Re-drafting the guidance to allow for greater discretion: This was a matter for the CoP. However, the Court indicated that the guidance should reflect what appeared to be regarded by the CoP as the existing position: a ‘common sense discretion not to record irrational complaints’ (at [123]).

In conclusion, at first instance and on appeal, each court identified different tests for evaluating the lawfulness of the HCOG: the former focused on whether the policy lead to an ‘unacceptable risk’ of unlawful conduct, and the latter on whether the policy ‘authorised or approved’ such conduct. 

It is an interesting feature of this case that the appellant succeeded under the more onerous standard of review. Whilst critics of Aand BF have argued that the effect of the Gillick test is to make it more difficult to challenge a policy issued by a public body, the outcome in Harry Miller shows that this will not always be the case. The Gillick test differs from the ‘unacceptable risk’ test by requiring a stricter focus on a comparison of ‘what the relevant law requires and what a policy statement says’ (A at [41]), and avoids statistical speculation to which courts are unsuited. However, as applied in Harry Miller, both forms of test operated as a gateway to the same substantive proportionality assessment. It was the difference in approach taken to that assessment that led to the reversal in outcome on appeal.

In other words, in cases of this type, it is the secondary proportionality analysis, rather than the gateway test, that is determinative of the underlying legal issue. And it is this aspect of Harry Miller, the substantive question of the appropriate balance to be struck between, on the one hand, the legitimate aims pursued by perception-based reporting across all five monitored strands of hate crime, and, on the other hand, the severity of the effects of such reporting on an individual’s Article 10 right to freedom of expression, that remains a subject of continuing controversy. It is likely to be one which soon returns to the courts.

With thanks to Professor Alison Young  for comments on a draft version of this post.

Autumn Ellis is a PhD candidate at Robinson College, University of Cambridge.  

(Suggested citation: A. Ellis, ‘Lawfulness of policies of public bodies and Freedom of Expression under Article 10 ECHR’, U.K. Const. L. Blog (28th February 2022) (available at https://ukconstitutionallaw.org/))