Same, Same but Different: Proportionality Assessments and Equality Norms

2.1 Introduction to Article 14

Article 14 of the ECHR states that ‘the enjoyment of the rights and freedoms set forth in this Convention shall be secured without discrimination on any ground such as sex, race, colour, language, religion, political or other opinion, national or social origin, association with a national minority, property, birth or other status’. It is clear from the treaty text and from the Court’s jurisprudence that Article 14 can only be invoked in connection with the other rights protected by the ECHR. A violation of Article 14 does not presuppose a breach of another right; it is sufficient that a practice interferes with an entitlement that falls within the ambit of a substantive right.⁶ Article 14 also provides an open-ended list of discrimination grounds or ‘statuses’. The Court has continued to add new statuses to the list, including, for example, age, sexual orientation, and disability.⁷ Some of these new statuses are quite different from the ones we typically associate with non-discrimination law: for instance ‘kind of prison sentence’, ‘place of residence’, and ‘former collaboration with secret services’.⁸ This has given rise to a controversy regarding whether Article 14 is primarily concerned with violations of the Aristotelian maxim that ‘like cases must be treated alike’ or whether its focus is on the social disadvantage experienced by certain groups (eg women, ethnic minorities, persons with disabilities, LGBTI individuals) irrespective of whether that disadvantage can be said to constitute differential treatment.⁹ As the following two sections illustrate, Article 14 prohibits both the differential treatment of people in similar situations and other practices that cause or reinforce social disadvantage. These sections also describe the basic rules for determining whether such practices can be justified.

2.2 Equal Treatment and Non-discrimination

The ECHR contains no definition of discrimination, but the ECtHR has developed a concept of discrimination in its jurisprudence. In short, the Court has established that, in order for an issue to arise under Article 14 of the ECHR, there must be a difference in the treatment of persons in analogous or relevantly similar situations on account of their sex, race, colour, or some other ‘status’.¹⁰ The claimant must show that he or she is in a situation that is analogous to that of others who are better off than the applicant. Two people or groups of people are, of course, never similarly situated simpliciter, but only similarly situated with respect to something. Thus, two persons can be similarly situated with regard to some domestic practices but not others. Citizens and migrants are, for example, similarly situated when it comes to the enjoyment of freedom of thought or the right to protection from racist attacks, but differently situated when it comes to the enjoyment of the right to vote in national elections or the right to benefit from certain welfare programmes.¹¹

In many cases, the Court does not discuss whether the claimant’s situation is in fact similar to that of others who are better off in relevant respects. Instead, such argumentation is postponed to the justification stage.¹² This sequencing is typically to the advantage of the claimant because, once the relevant similarity has been established, the burden of argumentation and proof shifts to the State, which must justify its practices. From the Court’s perspective, combining the discussion of relevant similarity with the discussion of the justifiability of the practice makes perfect sense in cases where the key problem with the practice under review is that the justification for it is weak. In such cases, the contextual reasoning necessary to establish whether two persons or groups are similarly situated with regard to a certain social good is very similar to the reasoning necessary to establish whether differential treatment is justified.¹³ This does not mean, however, that claimants can dispense with comparative argumentation altogether. The Court has rejected a significant number of discrimination claims on the basis that the claimant failed to demonstrate that he or she was situated in a similar way to others who enjoyed a protected freedom or social benefit that the claimant could not enjoy.¹⁴

Article 14 has been interpreted so as to cover both so-called direct discrimination (ie differential treatment based on a prohibited status) and indirect discrimination (ie disadvantageous treatment of protected groups resulting from the application of a general policy or measure that is couched in neutral terms).¹⁵ Violations of these prohibitions can arise out of actions as well as out of failures to act. Article 14 includes not only an obligation to refrain from adopting and implementing discriminatory laws and policies but also a number of positive obligations to mitigate the harmful consequences of past discrimination and to combat the discriminatory actions of private parties. In Horváth and Kiss v Hungary, for example, the Court held that States have ‘specific positive obligations to avoid the perpetuation of past discrimination’, including an obligation to ‘undo a history of racial segregation in special schools’.¹⁶ The Court has also declared that States have a duty to investigate, prosecute, and punish racist, homophobic, and religiously motivated violent acts committed by private individuals.¹⁷ The same obligation exists with regard to domestic violence, which disproportionately affects women.¹⁸ In relation to persons with disabilities, States must take reasonable action to remove the legal, physical, and other barriers that circumscribe the opportunities of such persons to enjoy their rights and freedoms on an equal basis with others.¹⁹ As this brief description of Article 14 illustrates, a wide range of state policies and practices may fall within the scope of this provision. Some of these practices serve legitimate aims and can be justified, as the next section explains.

2.3 Justified Practices

The ECtHR has on numerous occasions declared that not all disadvantageous treatment that is directly or indirectly linked to a discrimination ground violates the ECHR.²⁰ Policies and practices can be justified if they serve a ‘legitimate aim’ and if there is a ‘reasonable relationship of proportionality’ between the means employed and the aim to be achieved.²¹ Starting with the first criterion, aims are legitimate if they are recognised as such in the ECHR treaty text or are otherwise compatible with the treaty. This criterion serves to weed out policies with offensive aims (eg sexist, racist, or homophobic aims), but little else. In most cases, States claim that their policies serve respectable aims, and the ECtHR typically makes little effort to challenge such claims.²² In addition to the legitimate aims enumerated in Articles 8–11 of the ECHR, the Court has, for example, accepted public tranquillity,²³ the protection of ‘the family in the traditional sense’,²⁴ and ‘allocating a scarce resource fairly between different categories of claimants’²⁵ as legitimate aims within the context of non-discrimination.

The other part of the justification test includes a proportionality assessment. The policy under review must strike a fair balance between the protection of individual rights and freedoms and the collective interests of the community.²⁶ When assessing whether a fair balance has indeed been struck, a number of factors are taken into account, including the effects of the policy under review, the existence of alternative ways of achieving the policy’s aims, the basis for differentiation, and the social situations of those affected by the policy. Consideration is also given to whether the policy concerns a politically sensitive issue and whether there is a European consensus on what Article 14 requires in the circumstances at hand.

When it comes to the effects of the practice under review, the general idea is that it takes more to justify policies or practices that severely restrict the enjoyment of rights than it does practices with less severe consequences. Precisely how we ought to determine ‘how much’ a policy interferes with a specific right is seldom discussed in general terms, although some patterns can be inferred from the jurisprudence. Rigid or indiscriminate policies are often held to be more restrictive than policies that allow for exceptions.²⁷ Similarly, a policy that applies for a short period of time is often seen as less restrictive than a more permanent policy.²⁸

When discussing the effects of domestic practices, the Court also pays attention to alternative ways of achieving what the State is seeking to achieve. If there are ways of achieving the same aim that imply fewer restrictions on human rights and freedoms, then this speaks in favour of finding the policy under review disproportionate (this is sometimes referred to as the ‘least restrictive means’ rule). This has led the ECtHR to reject policies that are too far reaching. In Sidabras and Džiautas, for example, the Court rejected a policy that prevented former KGB agents from working not only within the public sector but also in a number of areas in the private sector for a period of ten years.²⁹ The policy had come into force nearly a decade after the collapse of the Soviet Union. Although the Court recognised the need to prevent persons who had served oppressive regimes in the past from taking up certain public positions, it did not see that this implied barring such persons from positions in the private sector as well.³⁰

The ECtHR has drawn on the American ‘suspect classifications’ doctrine as a way of signalling that certain forms of discrimination are more difficult to justify than others. According to this doctrine, differential treatment based on certain grounds, including gender/sex, ethnic origin/race, sexual orientation, and disability, is seen as inherently suspect and so as implying the need for a particularly persuasive justification if it is to be deemed lawful. According to the Court, it takes ‘very weighty reasons’ to justify differential treatment based on these grounds.³¹ Of course, negative attitudes towards certain groups or patriarchal ideas about the subordinate position of women can never justify the disadvantageous treatment of these groups.³² Nor is it sufficient that such differential treatment contributes to the aim in question. Rather, States must demonstrate that the differential treatment is necessary in order to achieve the aim of the policy.³³ The Court’s reasoning in Konstantin Markin is illustrative in this respect. The issue in the case was whether military servicemen could be refused parental leave when such leave was available to servicewomen. The stated aim of the policy was to secure the operational effectiveness of the armed forces. Whilst the Court recognised that it may be justifiable to deny certain categories of military personnel – male or female – parental leave on the grounds of their strategic position, their possession of rare technical qualifications, or their active involvement in military action, it maintained that a policy that denies servicemen parental leave on the basis of their sex violates Article 14 in combination with Article 8.³⁴ There are, however, situations in which differentiation based on a suspect ground has been accepted as lawful. For example, the Court still accepts as lawful domestic policies preventing same-sex couples from marrying and different-sex couples from registering their partnerships. It also accepts as lawful policies that exempt women (but not men) from the possibility of being sentenced to life imprisonment.³⁵

The Court has increasingly been paying attention to the situation of those affected by the policy in question. Some groups have been identified by the Court as particularly vulnerable and thus in need of extra protection or care from the government. These groups include Roma, persons with psychosocial disabilities, persons living with HIV, asylum seekers, persons belonging to the LGBTI community, victims of domestic violence, and women of colour working in the sex industry.³⁶ According to the Court, it takes very weighty reasons to justify restrictions on the fundamental rights of such groups.³⁷ States must be particularly careful about ensuring that their policies and practices are not based on stereotypical beliefs about the characteristics or behaviours of members of these groups, as this could perpetuate public prejudice and exacerbate social exclusion.³⁸

Last but not least, the ECtHR has established that States enjoy a certain amount of discretion (a margin of appreciation) when balancing the reasons for and against their policies in particular cases.³⁹ Because of their direct knowledge of local circumstances, traditions, and legal cultures, national authorities are ‘better placed’ than international courts to appreciate the values at stake and how they ought to be balanced in concrete cases.⁴⁰ The Court is particularly inclined to accept domestic regulations in policy areas that involve the balancing of politically sensitive interests, such as welfare policy or national security.⁴¹ Another factor that affects the degree of deference given to States is the presence or absence of a consensus among States Parties to the ECHR about what Article 14 requires in the situation under review. In situations where there is such a consensus, a policy that departs from this consensus is likely to be taken to violate the ECHR,⁴² although compelling reasons anchored in the specific political, historical, and legal context of the country in question may justify a departure from the common European approach to a specific issue.⁴³ With regard to issues on which there is no consensus, States have more room for manoeuvre. Indeed, the lack of consensus is an important reason why the Court still accepts domestic bans on same-sex marriage and regulations that ban the wearing of certain religious garments in public places.⁴⁴ The Court has not only used the existence or absence of a consensus on a certain issue in order to determine the degree to which it should defer to States when balancing equality interests with other domestic constitutional values and policy choices; it has also referred to such considerations in order to justify changes to its views about what Article 14 permits and prohibits. Changed attitudes towards parental leave and transsexuality, for example, have led the Court to change its position on fathers’ rights to parental leave and the legal recognition of the gender of post-operative transsexuals.⁴⁵

In sum, Article 14 prohibits both direct and indirect discrimination. Prima facie discrimination may be justified if the State practice in question serves a legitimate aim and is proportionate in relation to that aim. A plurality of factors is taken into account in determining whether a practice is proportionate, including the basis for differentiation, the aim and effects of the practice, alternative ways of achieving the practice’s aim, the social situation of those affected, and whether the practice concurs or conflicts with the European consensus on the issue (if there is one). However, it remains unclear how precisely these factors are supposed to relate to each other and influence the outcome of the balancing exercise implicit in the Court’s justification test. The Court’s reasoning on differential treatment based on ‘suspect’ grounds and its position on members of vulnerable groups imply that especially strong arguments are necessary to justify prima facie discrimination that is based on certain grounds or that affects persons that have suffered considerable discrimination in the past that has had lasting consequences. Likewise, the existence or absence of a European consensus on an issue affects the standard that must be met in discrimination cases. Still, these guidelines raise further questions that lack clear answers. One such question is whether ‘nationality’ (or ‘religion’) is a suspect ground and, if so, what this implies for the ability of States to restrict the access to certain rights and benefits to its own citizens.⁴⁶ In addition, it remains unclear how we ought to identify vulnerable groups, and this lack of clarity undercuts the usefulness of this doctrine.⁴⁷ And the Court’s application of the consensus rule raises questions about what qualifies as such a consensus. How similar must the relevant domestic interpretations of what the ECHR requires be, and how many States need to agree?⁴⁸ Moreover, these guidelines leave open the question of how to approach the not uncommon situation in which these different rules point to different conclusions, such as in cases involving discrimination based on suspect grounds in politically sensitive policy areas.⁴⁹ The next section explores whether Robert Alexy’s model of proportionality reasoning might provide a more structured approach to proportionality assessments in general and balancing in particular.

4.1 Introduction

As I mentioned above, Article 14 has been interpreted as expressing both the idea that, when it comes to the enjoyment of human rights, like cases must be treated alike and the idea that practices that cause or reinforce social disadvantage are unjust and must be opposed.⁶⁹ The first idea implies that Article 14 forbids arbitrary restrictions on rights and social goods on the basis of grounds linked to a protected status, interpreted in a broad sense. The second idea is less interested in whether the policy or practice constitutes same or differential treatment of similarly situated groups and more focused on the policy’s harmful effects on certain groups. That does not mean that the question of whether the policy makes arbitrary distinctions between persons in analogous situations is irrelevant; it just means that the absence of such arbitrariness is not sufficient to justify prima facie discrimination with particularly harmful effects. The following two sections discuss how these two ideas can be incorporated into the proportionality assessment, and into the balancing process in particular.

4.2 Incorporating the Idea of Treating Likes Alike into Proportionality Assessments

To incorporate the idea that like cases be treated alike into proportionality argumentation, the framework developed in this article adds two rules to Alexy’s model for proportionality assessments described in section 3. The first rule can be viewed as a specification of Alexy’s suitability criterion (ie State policies that interfere with human rights must contribute to their aims) tailored to the discrimination context. To be justified under Article 14, it is not sufficient for States to demonstrate that their policies contribute to their stated aims. If the policy constitutes direct discrimination, then the State Party must demonstrate that differential treatment on the basis of a protected ground serves the aim in question. This rule is omnipresent in the Court’s jurisprudence under Article 14, and it explains why the Court has found that there were violations of this provision in a number of cases involving so-called suspect as well as non-suspect grounds.⁷⁰ States may, for example, restrict maternity leave to women, but they may not restrict parental leave to women. Maternity leave aims to enable mothers to recover from childbirth and to breastfeed their babies if they so wish, whilst parental leave aims to enable a parent, ie the mother or the father, to stay at home and care for the child.⁷¹ A failure to meet this criterion arguably constitutes a sufficient reason to conclude that there has been a violation of Article 14. To hold otherwise would be tantamount to saying that differential treatment based on sex (or any other protected ground) may be justified even though such differentiation is irrelevant in achieving the aim and even though the policy makes no contribution to its purported aim.⁷²

In relation to indirect discrimination, a similar requirement could be formulated in order to oblige States to justify the disparate impact a particular policy has on a protected group. For example, if States decide to set up special classes to help pupils improve their command of the language of instruction, it is only to be expected that immigrant children and children from linguistic minorities will be overrepresented in such classes. It would be much harder, if not impossible, to explain, for example, an overrepresentation of pupils in wheelchairs, or, as in the case DH and Others v the Czech Republic, to explain why Roma children are overrepresented in special schools for children with intellectual disabilities.⁷³ We have no reason to believe that such disabilities are more common within the Roma population; rather, such a disparate impact suggests underlying bias and racism in the domestic system.

The second rule belongs to the balancing stage and seeks to ensure that the arguments put forward in defence of a particular policy must be internally consistent. A policy may seem proportionate when viewed in isolation but still be unjust when seen in context. It may, for example, seem reasonable to restrict the right to vote in national elections to citizens who are able to make judicious voting decisions.⁷⁴ The great value we attach to the democratic process and the need to protect the integrity of the electoral system could undergird such a claim. It is, however, also possible to argue that the right to vote is not dependent on the possession of cognitive skills such as the ability to critically assess political information. Voting is a subjective choice, linked to personal preferences, and it is not for the State to determine what constitutes a valid political opinion or which capacities are necessary to formulate such an opinion.⁷⁵ Even if both positions on the right to vote were legally tenable, which would imply that States had discretion on this matter, it seems wrong to accept that restrictions on voting rights for citizens with psychosocial disabilities could be justified on the basis of the first claim and, at the same time, to accept that the tolerance of uninformed voting by citizens without such disabilities could be justified on the basis of the second. Presumably, Article 14 requires that the reasons put forward to justify restrictions of rights apply to everyone.⁷⁶

This second rule differs from the first rule discussed above, which requires States to explain the rationale behind policies that stipulate differential treatment based on protected grounds or that have disparate impacts on protected groups. The problem with a policy that justifies restrictions on voting rights for persons with psychosocial disabilities in an inconsistent way is not that the State is acting on the basis of false or unsubstantiated beliefs about persons with psychosocial disabilities. A lack of certain cognitive skills certainly is more prevalent among members of this group than it is among persons without such disabilities. The problem, from the equal treatment perspective under discussion here, is that a policy that deprives persons with psychosocial disabilities of the right to vote is vastly overinclusive (many persons with psychosocial conditions are indeed able to make informed voting decisions) and that it holds them to a higher standard than many other groups. People with psychosocial disabilities are not the only ones whose ability to make judicious political choices might be questioned. A range of factors may impact our ability to make such choices, including whether we have consumed alcohol or drugs, whether we are very young (or, indeed, very old), or whether we are simply uninterested in politics. Whilst we accept age limits on voting, which can be justified by the fact that most children will eventually reach an age at which they will be allowed to vote, it would be unthinkable in most European democracies to require adult citizens without cognitive disabilities to demonstrate their knowledge of the political system to be able to cast a vote.⁷⁷ An internally consistent defence of a policy that required members of certain groups to pass a test before they could vote would have to engage with these questions and provide some other explanation of why it is reasonable to restrict voting rights for some citizens and not others.

4.3 The Harms of Unequal Treatment and Discrimination

Equality scholars describe the harms caused by discrimination in different ways, and, as several authors have noted, consensus on what these harms are remains elusive.⁷⁸ Having said that, a number of harms are typically highlighted by the ECtHR and equality scholars alike.⁷⁹ These include personal (emotional) harms and well as group-related ones. Starting with personal harms, empirical studies suggest that discriminatory practices have negative effects on the mental well-being of their victims. Persons who experience frequent exposure to discrimination and similar forms of unfair treatment report psychological distress, unhappiness, and depression.⁸⁰ The ECtHR has recognised such harmful effects of discriminatory practices. In DH and Others v the Czech Republic, for example, the Court held that the pupils placed in special schools for children with intellectual disabilities should be compensated for the anxiety, frustration, and humiliation they had suffered as a result of that policy.⁸¹

The group-related harms of unequal treatment and discrimination take a variety of forms and have been conceptualised in numerous ways. Denise Réaume and Deborah Hellman claim that what is harmful about discrimination is that it is an affront to human dignity: discrimination conveys a demeaning message that some are of lesser worth than others.⁸² Charilaos Nikolaidis argues that social oppression, in the form of stereotyping and stigmatisation, distinguishes violations of Article 14 from other human rights violations.⁸³ Other authors have described the harms of discrimination in terms of uneven or unfair distribution of material resources, political power, and opportunities to pursue one’s life projects and similar social ‘goods’.⁸⁴ In the recent literature, pluralistic accounts have recognised that different forms of discrimination cause different harms and combinations of harms.⁸⁵

The Court has not engaged in a principled discussion about the specific harm(s) brought about by domestic practices that violate Article 14 but has approached such questions on a case-by-case basis. Racial and homophobic violence, for example, have been said to disrespect human dignity.⁸⁶ The harm brought about by the segregated education of Roma children was, by contrast, described not as a violation of human dignity but primarily in terms of substandard education, which, in turn, was described as limiting the children’s future career opportunities and their capacity to participate actively in mainstream society.⁸⁷ The harm caused by stereotyping has primarily been discussed in relation to policies that reinforce stereotypical ideas about men and women and about vulnerable groups. In Konstantin Markin, for example, the Court held that policies that reinforce the stereotype of the male breadwinner and the female homemaker are harmful to women’s careers as well as to men’s family lives.⁸⁸ In most cases, however, the Court’s discussion of the harm caused by a policy focuses on whether the claimant has been placed in a position of comparative disadvantage, that is, whether he or she cannot enjoy a right, or an aspect of a right, set out in the Convention to the same extent that others can.

The harms of discrimination and unequal treatment not only take various forms but also differ in the impact they have on people’s lives. Some discriminatory practices have profound consequences. The far-reaching and long-lasting effects of segregated education have already been mentioned. Systems of plenary guardianship not only prevent certain persons with intellectual and psychosocial disabilities from voting, and thereby ensure that their interests remain low on the political agenda, but also deprive them of the opportunity to make almost every important decision in life, including decisions about where and how to live.⁸⁹ The failures to address gender-based, racist, homophobic, and similar forms of violence are other examples of discrimination with particularly severe implications. Some other practices that fall within the ambit of Article 14 are clearly unlawful but have a more modest impact on people’s lives. Examples of such practices would be laws that prevent married couples from using the wife’s (rather than the husband’s) surname as the family name, or laws that mean that the burden of jury service is borne disproportionately by men.⁹⁰ This is not to trivialise the injustice of such discriminatory practices or to deny their role in sustaining the unequal power relationship undergirding more brutal forms of discrimination, such as gender-based violence. It is, however, to recognise the possibility of moral and legal differentiation. Alexy’s balancing model asks us to assess ‘how much’ a particular policy or practice interferes with the right(s) at stake (here Article 14).⁹¹ The magnitude of the harm brought about by the policy in question is a key factor in such an assessment, and it thereby affects how strong or weighty the reasons must be in order to justify the policy in the circumstances at hand.

The magnitude of the harm caused by a particular policy also depends on the situation of those affected by it. Discriminatory policies affecting already marginalised groups, and thereby compounding the hardship experienced by members of these groups, arguably generate more harm than policies which affect more privileged groups. This may explain the Court’s position that it takes particularly weighty reasons to justify prima facie discrimination against vulnerable groups and that States must avoid perpetuating stereotypical beliefs about members of these groups. Another reason to view policies that reinforce negative stereotypes as particularly harmful is that stereotypes affect all of us and serve to maintain existing power relationships between privileged and marginalised groups, which makes it more difficult to create equal and inclusive societies.⁹²

4.4 A Framework for Proportionality Argumentation within Discrimination Analysis

This section summarises the discussion in the preceding two sections in order to outline a framework for evaluating justifications of prima facie discrimination under Article 14 of the ECHR. The evaluation commences with an assessment of the legitimacy of the policy’s aim(s) and the means–ends rationality of the policy. This initial examination can be viewed as a simple way of rejecting at the outset policies and practices that are clearly unjustified. The following criteria must be met:

As discussed above, legitimate aims are aims that are compatible with the ECHR.⁹³ This weeds out policies with particularly offensive aims, such as racial segregation, but this criterion is a low threshold. The second criterion ensures that policies do not make distinctions between groups for irrelevant reasons.⁹⁴ The third criterion prohibits policies that are more restrictive than is necessary to contribute to their aims.

To determine whether a particular regime is proportionate (in the narrow sense), the reasons speaking in favour of its compliance with the ECHR must be weighed against the reasons speaking against its compliance. To count as lawful, the following criteria must be met:

These criteria describe a balancing process that is specifically tailored to the discrimination context. As outlined in section 3, balancing is a form of legal argumentation that obliges decision-makers to take all legally relevant considerations into account. These include arguments about the scope and content of the legal norms at stake, about the ‘costs’ of accepting or rejecting the domestic practice under review, and about the reliability of the assumptions upon which these arguments are based.⁹⁵ As I have explained, the harms of unequal treatment and discrimination include both personal and group-related harms, and some practices are worse than others because of the magnitude of the harm they cause. This explains why it takes weightier reasons to justify such practices. To determine the weight of the reasons in favour of a policy, we must consider the costs of amending it in such a way as to produce less harm. Needless to say, it is difficult to determine such weights, and there will often be disagreement about them. Such disagreement is clear from the different ways that States Parties to the ECHR approach certain matters (eg same-sex marriage) and from the many dissenting opinions accompanying the Court’s judgments in discrimination cases. Few, however, would contest that it makes legal sense to say that some policy changes interfere more with States’ legitimate interests than others. The costs of modifying policies that fail to meet the second criterion outlined above, and therefore do not contribute to their aims, are certainly minimal. The costs of accepting different expressions of human diversity (eg the wearing of the hijab in public places) are arguably also low, unless they interfere with the rights of others.⁹⁶ Transforming our societies into inclusive ones where everyone has a genuinely equal opportunity to participate in public affairs and benefit from education, welfare programmes, and other social goods is, on the other hand, a project with more far-reaching implications for State budgets and domestic legislators’ room for political manoeuvre.⁹⁷ These implications may nevertheless be ‘outweighed’ by the devastating consequences of social exclusion and of the failure to accommodate the needs of marginalised groups, as in the case of the segregated education of Roma children.⁹⁸

The fifth criterion serves to ensure that any justification of unequal treatment is internally coherent. The content of this criterion has been discussed in section 4.2, and I will not repeat this discussion here.