Religious freedom and the right against religious discrimination: Democracy as the missing link

The case for a democratic framework to protect religious interests

The starting point of the democratic framework is that religious interests are endowed with a dual dimension. They are important as a negative liberty, to protect believers from intrusions and interferences, as well as express a positive value to society, to support pluralism and equality. This premise is also that of the ECtHR, which, in one of its very first cases involving article 9 ECHR, stated that:

Freedom of thought, conscience and religion is one of the foundations of a ‘democratic society’ within the meaning of the Convention. It is, in its religious dimension, one of the most vital elements that go to make up the identity of believers and their conception of life, but it is also a precious asset for atheists, agnostics, sceptics and the unconcerned. The pluralism indissociable from a democratic society, which has been dearly won over the centuries, depends on it. ¹²

Through the pluralism they foster, religious interests thus not only matter for their holders but for democracy itself. By encouraging deliberation and reason-giving, a democratic approach to the protection of legal interests would help, in the words of Cohen, ¹³ ‘to constitute a political community of equals’ and ‘enable co-deliberation among political equals on a terrain of public reason’. In this deliberative discursive sense about the terms of legitimacy, religious interests would have an enhancing effect on democracy. This democratic dimension simultaneously reveals the positive value of religious freedom and explains that religious requests come with limits attached. This inherent tension carries consequences both for courts and for citizens. First, the democratic approach thus refutes the proposal that religious requests should receive priority in adjudication. ¹⁴ Secondly, all citizens, whether religious or not, would need to accept the horizon of a democratic constitutional overarching framework, and the obligations of reciprocity and renewal it carries. Illiberal religious (or non-religious) citizens who would deny the possibility and reasonableness of views contrary to their own would therefore incur limits. I have demonstrated elsewhere in full how such a deliberative democratic framework could support the dual dimension of religious interests and provide guidelines to judges confronted with delicate and novel controversies. ¹⁵ Suffice to say here that the democratic framework yields three criteria for judicial reasoning: a method of avoidance, a principle of inclusion and a principle of revision. Under the method of avoidance, judges are to refrain from interfering with religious beliefs. The principle of inclusion, the second feature in the democratic framework, will, however, trigger intervention by courts to redress inequalities. Finally, the principle of revision expresses the notion of reciprocity and requires that citizens review their commitments, in light of the overall political framework and that, vice versa, state institutions review positions and legal solutions as new contestations emerge. The democratic framework thus requires that the terms of legitimacy of our common vivre ensemble are constantly subject to revision and that the revision is conducted in a way that respects citizens’ autonomy (hence the method of avoidance) and citizens’ equality (hence the principle of inclusion).

More concretely, under the democratic approach, both the method of avoidance and principle of inclusion combine to dictate that courts do not determine in advance which religious practices amount to acceptable manifestations but take on board the claimant’s interest in expressing her religious faith, however conspicuous or inconsistent the manifestation may appear at first sight. This does not entail that courts cannot legitimately set limits to religious requests, but that no pre-selection or filters should be put in place at the outset. The positive value which the democratic approach confers upon religious voices forbids their outright exclusion, through abstract preliminary filters and distinctions. The importance of religious interests for democracy precisely lies at the heart of the decision by the ECtHR in Eweida ¹⁶ to abandon the ‘contracting out’ or ‘specific situation rule’ approach, whereby employees could be deemed to have contracted their religious rights out of the workplace, either by agreeing to explicit contrary requirements in their contract of employment, or, implicitly, by subjecting themselves to workplace regulations and policies. Such abstract neutralisation of religious voices in the workplace not only placed an undue burden on religious employees, forced to choose between their religious and professional commitments, but impoverished the pluralism within the workplace. Besides, where preliminary filters draw a priori distinctions between types of religious symbols or religious practices, they betray the method of avoidance, as courts and legislators then unilaterally decide in advance which symbols and practices may qualify as religious protected practices. The method of avoidance I promote however does not equate to a retreat from judicial review; if courts should not unilaterally impose their own definition of religion, they should nonetheless, as will be apparent in the analysis of cases below, assess the factual complex context in which the religious request is raised. ¹⁷ If judges simply adopted a position of retreat every time religious freedom is involved, they would potentially infringe the principle of inclusion, as holders of competing rights might then see their own interests excluded outright. By shifting the controversies out of courtroom, they would also deny the opportunity for right-holders to engage with one another’s conflicting positions. Seen through the overarching goal of a common political framework construed in light of the principles of revision and inclusion, the method of avoidance is therefore compatible with judicial review of controversies over religious freedom. The method of avoidance under the democratic approach does not purport to safeguard spheres of religious autonomy, immune from judicial scrutiny, but to ensure that no interferences hinder religious (and non-religious) voices from participating fully in public debate. While the principle of revision postulates that one cannot and should not identify in advance one single legitimate outcome in each given case, the democratic approach will therefore insist that the chosen outcome derives from a thorough scrutiny of the competing interests at play, as detailed in the final section.

In this article, I aim to underline features of my democratic approach, which have implications for religious discrimination law. Many authors have acknowledged a growing trend towards a reduction in the gap between the religious belief reasoning by the ECtHR and the religious discrimination reasoning by the CJEU. ¹⁸ Yet, in recent cases discussed below, the CJEU seems to have reintroduced a divide. In the section to follow, I will argue, based on the proposed democratic approach, that the re-emerging gap, as reflected in recent CJEU rulings, is to be resisted as it undermines the inclusiveness of religious voices in the workplace.

For a greater inclusiveness of religious voices under indirect discrimination law

In this section, I build upon the democratic framework to criticise two features of indirect discrimination law case-law and the underlying idea they reflect that discrimination law, unlike religious freedom, should be subject to a higher threshold. ¹⁹ First, indirect discrimination law has been construed in recent CJEU cases as a tool to free the workplace from religious symbols, save for limited exceptions. Secondly, protection under indirect discrimination law has at times been subject to the claimant proving membership to a disadvantaged vulnerable group. As I will show in this section, both these features undermine the pluralism which the democratic approach seeks to promote.

Deliberative inclusion v. abstract pre-determination

Under the proposed democratic framework, the emphasis on open-ended deliberation would guard against the temptation, present in some of the case-law on indirect discrimination, to revert to strict and abstract lines of demarcation between acceptable and unacceptable signs of religious affiliation in the public sphere. In Achbita, the CJEU, following the Opinion of AG Kokott, ²⁰ held that abstract rules could be allowed under company neutrality policies, provided employers enforced them consistently and proportionally. ²¹ To establish the legitimacy of the neutrality policy, the Court merely stated that ‘an employer’s desire to project an image of neutrality towards both its public and private sector customers is legitimate, notably where the only workers involved are those who come into contact with customers’. The statement however fails to explain what commercial interests the employer (a security company) might have in projecting an image of neutrality towards religion. The reference to contact-facing role moreover seems to contradict the court’s concomitant ruling, in the case of Bougnaoui, ²² in which the Court excluded customers’ wishes as justification for a direct discriminatory measure. It is not clear why the fact that these customers’ objections had been anticipated and entrenched into a neutrality company policy should drastically alter the assessment. The same focus on abstract pre-determination characterises the assessment of proportionality requirements. According to the Opinion of AG Rantos in the subsequent cases of C-804/18 IX v Wabe and C-341/19 MH Müller, ²³ a neutrality policy would satisfy proportionality requirements if it only banned large-scale religious signs. Such pre-demarcation lines between acceptable and unacceptable interests betrays deliberativeness as from the outset, certain interests, deemed to lead to ostentatious practices, are excluded outright. The fact that these clear-bright rules rely on a company neutrality policy is of little relevance. If the Directive is to offer a work environment free of prejudice; if, under my democratic approach, religious interests have a positive value for public debate and society as a whole, the existence of a company policy should not per se be sufficient to defeat that aim and exclude outright certain religious voices.

More specifically, the reasoning adopted in Achbita, and by AG Rantos’ Opinion in Wabe and Mueller, infringes the three abovementioned criteria of avoidance, inclusion and revision within the democratic approach. It does not respect the method of avoidance as courts are led to directly and unilaterally assess the legitimacy of religious practices, and dismiss those which are too visible. It goes against the principle of inclusion, as mainstream religious beliefs are less likely to be affected than minority practices. It, finally, falls foul of the principle of revision, as dialogue and engagement between protagonists is skewed by the greater weight granted at the outset to employers’ decisions. Albeit superficially justified for the sake of equality and the smooth conduct of business, the outcome in Achbita is problematic as it conflates equality with consistency and harmony in the workplace with neutrality. The fact that the policy applied consistently to religious convictions as well as philosophical and political beliefs weighed heavily in favour of the employer. Indeed, from a formal egalitarian perspective, this consistency satisfied the requirement of equal treatment between employees. However, under a democratic approach, such emphasis on consistency does not suffice. Beyond the procedural requirement of a consistent enforcement and neutral phrasing of the policy, a democratic approach requires the employer to justify objectively why and to what extent the restrictions to religious freedoms might be necessary. Consistency moves the gaze to implementation issues without having first verified that a neutrality policy was legitimate in the first place. Contrary to the presumption of legitimacy which the Court in Achbita grants to the company policy, the democratic approach therefore avoids that some members of society see their access to employment and their (religious) voice excluded from the workplace at the outset. The same rationale explains why membership to a vulnerable group, at times required as a threshold to a discrimination law claim, falls foul of the democratic approach.

Democratic approach and group requirement under indirect discrimination law

Authors differ widely as to the normative foundation of discrimination law, ²⁴ but there is consensus as to the inclusiveness goal/consequence of discrimination law towards vulnerable categories. ²⁵ This focus on vulnerable disadvantaged groups under discrimination law has at times led authors to argue that membership to a disadvantaged group is a key requirement for a claim under religious discrimination law. ²⁶ In the Eweida case for example, ²⁷ Ms Eweida had complained that the refusal by her employer, British Airways (BA), to allow her to wear a visible cross on top of her uniform amounted to discrimination on the basis of religion. The Court of Appeal dismissed her claim that BA’s uniform policy constituted indirect discrimination and held that even if it did, it was justified. Sedley LJ held that the Directive did not intend ‘that solitary disadvantage should be sufficient’, ²⁸ thereby endorsing the finding by the Employment Appeal Tribunal (EAT), that the claimant must prove that the inconvenience she had personally suffered had caused a disadvantage to a group. ²⁹ To establish the absence of a group disadvantage in the case, the Court pointed to the peculiarity of the claimant’s request both within the workplace and within Christianity. ³⁰ Compared to the views of the other 30,000 members of BA staff and other fellow Christians, Ms Eweida’s request seemed to be ‘a personal choice rather than a religious requirement’. ³¹ Whilst the personal nature of Ms Eweida’s request makes it impossible for her employer to anticipate, it does not explain why, once the employer becomes aware of it, refusal of any accommodation would be ipso facto legitimate. The connection between the group disadvantage requirement and the justification of indirect discrimination is not therefore as self-evident as is sometimes assumed. ³² Moreover, the underlying concept of vulnerability is itself fraught with difficulties.

Vulnerability: An ambiguous concept

In many cases involving religious interests, identifying which of the parties belongs to a (more) vulnerable group will be delicate. Recent cases involving religious interests have given rise to ‘a clash of vulnerabilities’, with different protected characteristics pitted against one another. Take the Ladele case for example, ³³ in which a Christian registrar refused, on religious grounds, to take part in the celebration of same-sex unions. As McCrudden puts it, ³⁴

On one level of the analysis, we might say that it is clear that Christians in England are much less subordinated than those who are gay (…) Ms Ladele was a Christian, but she was also black, a member of a small Evangelical Church, working in Islington Council, which is amongst the most LGBT-friendly local authorities in London, a city that is among the most secular in the world.

Outside of clashing claims of vulnerabilities, the requirement of a group disadvantage will in itself often be open to challenge. Fineman’s work warns us against the simplistic dichotomy between advantaged and disadvantaged groups: ‘The most pernicious effect of the segmenting of a general population so that only some are designated as vulnerable … is that such segmentation suggests that the rest of us are not vulnerable’. ³⁵

Finally, even assuming that it may be possible to identify deprived categories, the requirement of membership to a vulnerable group serves less of a purpose in a democratic deliberative sense, as a means to protect and promote the pluralism of views in public life. Religious interests also express a worldview, and from that perspective, there is no reason to presume that certain worldviews do not need legal protection because of the groups from which they emanate. ³⁶ Nor can we square the circle by neatly separating the identity and worldview dimensions of religious interests into the discrimination and religious belief categories, respectively. As the Eweida case illustrates, the two aspects are tightly interwoven. Ms Eweida’s request to wear a Christian cross at work involved both a fundamental worldview and an identity claim. Denying her request had an adverse impact on both aspects: it both reduced the expression of views in the workplace and impaired access to salaried employment. It both undermined her identity and reduced the pluralism of views and beliefs allowed in the democratic microcosm of the workplace. Splitting the two dimensions of religious interests would therefore be artificial and distortive of the reality of adjudication (where both legal bases overlap) and of the lived experience of claimants. ³⁷ The democratic foundation of religious interests therefore usefully insists that religious interests be recognised (although not necessarily upheld) under discrimination law beyond egalitarian concerns. In so doing, the democratic approach will also avoid an impoverishment of the diversity of views and practices allowed in the workplace and broader public life. Inevitably, the diversity that the democratic approach brings into the workplace and beyond, into public life, will at times highlight conflict. In the section to follow, I will examine, through recent controversies, how the democratic approach might guide judicial balancing tests used to arbitrate between competing interests.

Revisiting balancing tests under discrimination law

In this section, I draw upon the democratic approach to make a plea for stricter balancing tests in cases involving religious freedom controversies, including when they are raised under discrimination law. While strict majoritarian parliamentary conceptions of democracy, which locate all democratic decision-making with elected members of Parliament, might object to the greater discretion potentially afforded to courts through balancing tests, ³⁸ the leeway of courts to review in concreto the fairness of balancing considerations framed by Parliament is compatible with constitutional views of democracy, which emphasise checks and balances. ³⁹ More positively, the deliberativeness qualities associated with democracy ⁴⁰ can only be enhanced if courts, through such balancing exercises, are able to offer a forum for competing interests to meet. ⁴¹ In that deliberative light, courts are not usurping powers from elected members of Parliament, but by putting to the forefront the competing interests underlying legal decisions, are taking part in the democratic process. ⁴² Judicial balancing tests can ensure that all interests underlying a particular dispute be given a priori consideration and avoid that some members of society see their access to basic services and domains such as employment and their (religious) voice excluded at the outset from certain spheres (such as the workplace). Moreover, importantly, proportionality tests avoid an impoverishment of the diversity of views and practices allowed in the workplace and broader public life, unless justified by a legitimate aim and proportionate to that aim. The emphasis on balancing tests thus matches the dual dimension conferred to religious interests under the democratic framework; it simultaneously protects the interests of religious citizens and enriches public life. In this section, I will show that there is no inherent obstacle in adopting a balancing approach under discrimination law, before throwing some light on the implementation of contextual proportionality tests under my proposed democratic framework.

Contextual proportionality tests: A possibility under discrimination law

Writing about recent decisions under discrimination law, Julian Rivers ⁴³ observes that: ‘equality law is not simply the expansionist vehicle for a post-liberal conception of equality, or indeed of any particular conception. Roughly speaking, and with inevitable imperfections, it seeks to preserve a balance of constitutional rights’. This balancing dimension of non-discrimination laws has been more apparent in recent case-law. In Egenberger ⁴⁴ and IR, ⁴⁵ for example, the CJEU insisted that equality laws entailed a confrontation between competing interests: the religious autonomy granted to churches could not therefore deprive individuals of their right to judicial review. ⁴⁶ In Egenberger, a job applicant applied to the German Evangelisches Werk for a post, which consisted of drafting a report on the United Nations International Convention on the Elimination of All Forms of Racial Discrimination. She complained that her lack of religious beliefs was the only reason she had been turned down and that she had therefore been victim of discrimination on the ground of religion (or lack of). In response, the employer argued that the requirement of Church membership was a genuine occupational requirement under Article 4(2) of the Directive, which allows employers to impose restrictions upon employees’ rights for the sake of a (religious) work ethos. The question raised was whether the determination of what amounts to such genuine, legitimate and justified occupational requirement within the meaning of Article 4(2) could be left to the religious employer. The CJEU clearly answers that it could not. The strict judicial scrutiny that follows undeniably impinges upon the employer’s own assessment of the scope of his own religious requirements. Let me explain why such judicial review is however compatible with the method of avoidance under my democratic approach.

The democratic approach seeks to promote religious interests in their dual dimension as negative liberties and rights to participation. Each of the three criteria under the democratic approach reveals and guarantees that religious freedoms are important not only for believers themselves but also for democracy. By protecting the equality between citizens, the principle of inclusion also preserves the enriching pluralism that religious diversity brings to public debate. The principle of revision guarantees that limits set to religious freedom and competing rights can evolve and be constantly reviewed through dialogue and exchange, which is also enriching to public debate. Finally, the method of avoidance, whilst protecting religious citizens and communities from state encroachments upon matters of religious doctrines, will also serve the horizon of a shared common space. The method of avoidance will therefore preclude courts from engaging with issues of religious scholastics, which would both interfere with religious beliefs and clutter democratic debate. It will not however prevent courts from examining how the religious claim may affect competing rights. If it did, certain spheres would potentially be exempt from principles of equality, fairness and inclusion and the overarching goal of a vivre ensemble respectful of pluralism would be lost. The insistence of the Court on a strict proportionality test in Egenberger is therefore perfectly in line with the democratic approach. It is also a feature of convergence between the religious discrimination and the religious freedom routes.

Just as the CJEU concluded in Egenberger that a strict judicial review and a proportionality test by the national courts was necessary to decide between conflicting interests, ⁴⁷ the Grand Chamber of the ECtHR similarly asserted, in Fernández Martínez, that:

a mere allegation by a religious community that there is an actual or potential threat to its autonomy is not sufficient to render any interference with its members’ rights to respect for their private or family life compatible with Article 8 of the Convention. (…) The national courts must [conduct] an in-depth examination of the circumstances of the case and a thorough balancing exercise between the competing interests at stake. ⁴⁸

The reasoning of both European Courts as regards justification of interferences with and restrictions to religious interests is therefore strikingly similar, despite their differing legal basis. This convergence proves that there is no inherent obstacle to balancing exercises under discrimination law. ⁴⁹ It thus refutes the notion that direct discrimination law would be averse to balancing tests. As Hugh Collins rightly puts it, ‘it is evident that some general defences such as an occupational requirement embody a particularised statement of the test of proportionality’. ⁵⁰

Such balancing extends beyond the occupational requirement justification. In the case of Bull v Hall, ⁵¹ the House of Lords (as it was then) also felt the need to enquire whether the restrictions imposed by anti-discrimination law provisions upon Christian Bed & Breakfast owners were justified. This justification process would prove, according to Hugh Collins, that direct discrimination law does not ignore competing interests. ⁵² The way in which courts decide to draw the contours of direct discriminatory practices inevitably includes an examination of the alleged discriminator’s own conflicting interests. In the Ashers Bakery case, ⁵³ for example, the UK Supreme Court held that a baker’s refusal to make a cake decorated with the words ‘Support Gay Marriage’ did not amount to direct discrimination on the ground of sexual orientation. The objection was to the message (the inscription, which was to decorate the cake) rather than to the messenger, the gay customer who had placed the order. The baker would have refused to inscribe the said message on all cakes, regardless of who had ordered it. Interestingly, it is to be noted however that the Supreme Court in Ashers did not characterise the tension at play as involving equality versus religious interests, but as one between freedom of speech and political discrimination. ⁵⁴ Beyond the particular legal circumstances of the case, ⁵⁵ the chosen characterisation may signal an avoidance strategy, designed to avoid fuelling ‘a culture war’ surrounding religion. In so much as such an approach helps set the tone for an appeased confrontation of arguments at a later stage, it fits in with the deliberativeness of the democratic approach. The importance granted in Ashers on freedom of speech, which would be curtailed if the baker were forced to endorse a message he objected to, is also conducive to the pluralism of views sought under the democratic approach. In Bull, the House of Lords refrained from following two opposite paths, which would both have betrayed the democratic approach. One would have consisted in undermining the weight of religious objections, merely because of the commercial context in which they were raised: the enriching pluralism that religious interests bring would be seriously hindered if religion were carved out a priori from key sectors of public life. The other would have been to undermine the infringement to equality interests by pointing to the possibility for aggrieved couples to find an alternative provider. Relegating equality interests to accessing services would not satisfy the democratic aspiration for a vivre ensemble respectful of diversity.

The increasing convergence in the case-law, under both the religious freedom and the religious discrimination routes, towards proportionality tests therefore matches the emphasis under the democratic framework on contextual deliberations. However, proportionality tests will only comply with the democratic framework if they are implemented in a way which bolsters deliberativeness and open-endedness.

The features of contextual balancing assessments under the democratic framework

In the last part of this section, I will analyse the ritual slaughter ruling of the CJEU ⁵⁶ and point out where, arguably, the court has failed to carry the balancing test satisfactorily in light of the democratic framework. Whilst carrying out a balancing exercise between the competing interests at stake, the CJEU did not, as I will show, abide by the three abovementioned criteria of avoidance, inclusion and revision. The CJEU considered that the Flemish decree, which abolished the previous derogation from prior stunning before the slaughter of animals, ⁵⁷ was compatible with the religious exemption provided for under Regulation 2009 (EC) No. 1099/2009 of 24 September 2009 (on the protection of animals at the time of killing (thereafter the Regulation). ⁵⁸ Unlike AG Hogan, ⁵⁹ who had considered that the decree was seeking to remove the exemption granted on religious grounds in the Regulation from the requirement of prior stunning altogether, ⁶⁰ the CJEU rules that the decree leaves the core of the exemption intact. ⁶¹ Since the death of the animal would still result from bleeding (as required under religious rites) rather than from the prior stunning, ⁶² the decree, according to the CJEU, would not erase the religious exemption provided for in the Regulation but merely narrow it down. Whilst the court adopts a measured and evidence-based approach, its unilateral interpretation of religious requirements betrays the method of avoidance under the democratic framework, as the state is led to interfere unilaterally with matters of religious doctrines. It is to be feared that religious communities might find this interpretation of their own religious requirements by secular authorities patronising.

Moreover, if the CJEU insists in this ruling on the need to adopt a balanced approach, it considers that such balanced outcome has already been reached outside of religious communities, within the Flemish decree. Such a non-discursive way of balancing competing interests would not comply with the principle of inclusion, the second feature under the democratic approach. Minority voices under the democratic approach need not only be taken into account, but actually need to be heard. Finally, the ruling also stands at odds with the principle of revision, the third criteria under the democratic approach. The court indeed notes that scientific consensus would now present prior stunning as the optimal means of reducing the animal’s suffering at the time of killing. ⁶³ Such observation is likely to stigmatise religious communities (presented to be at odds with science), create backlash and trigger enforcement problems. Contrary to the principle of revision, it also restricts the need for reviewing the outcome, since science has already sealed the debate. Only new scientific evidence could justify reopening the debate.

Whilst signalling a more contextual, hence, according to my democratic framework, a more welcome approach than the previous Achbita ruling, ⁶⁴ the ritual slaughter ruling would therefore still lack in deliberativeness. In comparison, AG Hogan’s opinion seemed more in line with the democratic framework I promote. The Opinion was compatible with the three criteria at the core of the democratic approach. It respected the method of avoidance by preventing the state from unilaterally prescribing the manner, in which a religious rite was to be carried out ⁶⁵ ; it followed the principle of inclusion by dismissing a solution, which would exclusively affect minority religious communities. It was, finally, in line with the principle of revision by leaving open the possibility of further restrictions and encouraging engagement with religious communities to improve animal welfare, without stigmatising religious communities as being inherently insensitive towards the interests of animals. ⁶⁶

In this section, I have argued that contextual balancing exercises between competing interests should (and to a large extent have) become a key unifying feature of both the religious freedom and the religious discrimination routes. Having explained how the traditional reluctance against such balancing tests under discrimination law has already largely been overcome and contradicted by the case-law, I have shown how the three criteria at the heart of the democratic approach (method of avoidance, principle of inclusion and revision) can avoid implementations of the balancing tests which fall short of inclusiveness and deliberativeness, as was, I have argued, the case in the CJEU ritual slaughter ruling.