Introduction
Equality stands at the center of many contemporary political debates. On the one hand, there is a perceived rise in social inequality.Footnote 1 This has not only led to a controversial political debate about how to address such inequalityFootnote 2 but also to a fierce discussion over whether this observation is indeed empirically robust.Footnote 3 On the other hand, there is an equally intense debate over the vulnerability of certain social groups and adequate means to protect such groups against discrimination. The Black Lives Matter movement, calling out continuing racial discrimination,Footnote 4 and the #MeToo debate, highlighting discrimination against and the sexual harassment of women, have not only drawn attention to existing social injustices but also provoked considerable political backlash.
Considering the pitfalls of political debates around equality, courts must tread carefully when they operationalize the meaning of equality. In this article, I argue that they have found different solutions to this challenge. The article identifies four distinct models of equality which diverge quite significantly.Footnote 5 These models are the equality-as-non-discrimination model, the equality-as-reasonableness model, the positive-equality model and the deferential approach to equality. I argue that none of these models has an absolute normative advantage over its alternatives. Instead, each has different normative strengths and weaknesses. Which model fits best depends on the societal and institutional context in which the court operates.
The article is largely a conceptual paper that describes the different models and discusses their normative strengths and weaknesses. However, it also has an empirical dimension, arguing that the four models can be observed in judicial practice. It draws on examples from a range of jurisdictions.Footnote 6 The assessments of the various jurisdictions discussed are based on an analysis that formed part of a larger comparative project focusing on the systematic examination of the constitutional equality case law of the jurisdictions under review.Footnote 7 The article proceeds in three steps. First, it examines the forms of equality norms in domestic constitutions and international human rights treaties. Second, the article recapitulates the international theoretical debate about the meaning of equality. The theoretical disagreement over how best to conceptualize equality gives courts significant leeway when they operationalize the concept. This is analyzed more closely in the third part, which introduces the various approaches to equality and discusses their normative strengths and weaknesses. The final part then advances the argument that there is no one-size-fits-all model for equality. Instead, the best approach depends on the cultural and institutional context in which a court operates.
Equality norms in a comparative perspective
Most domestic constitutions and international human rights treaties contain equality norms. The wording of these equality clauses is fairly similar, and they usually come in three different forms.Footnote 8 Some equality provisions are formulated as general equality norms, stipulating that everyone is equal before the law. Examples are the Equal Protection Clause of the Fourteenth Amendment to the U.S. Constitution or Article 24 of the American Convention on Human Rights (ACHR).
Another set of norms is framed as non-discrimination provisions. These protect against distinctions based on certain specifically listed suspect grounds. Some non-discrimination clauses contain exhaustive lists, such as Section 21 of the New Zealand Human Rights Act (in conjunction with Section 19 of the New Zealand Bill of Rights Act), while most include open-ended lists in which the enumerated grounds are merely illustrative. Examples of the latter include Article 14 of the European Convention on Human Rights (ECHR) and Article 3.IV of the 1988 Brazilian Constitution. Finally, many individual rights catalogs combine a general equality guarantee with a non-discrimination guarantee. Examples include Section 15 of the Canadian Charter of Rights and Freedoms, Article 3 of the German Constitution, Article 12 of the Constitution of Singapore and Article 26 of the International Covenant on Civil and Political Rights.
The specific form of an equality norm does not necessarily determine the equality doctrine that a court will apply. A key example of this is the jurisprudence of the U.S. Supreme Court.Footnote 9 Through its use of the tiered scrutiny framework, which applies heightened scrutiny to certain suspect classifications, the Court interprets the Equal Protection Clause as a non-discrimination provision. At the same time, non-discrimination norms do not preclude the application of a more stringent standard of review, even in the absence of a suspect classification. As noted, many such provisions contain open-ended lists that allow courts to address distinctions not explicitly enumerated.
The debate on the meaning of equality
While the language of equality norms in domestic constitutions and international human rights treaties is strikingly similar, the theoretical debate over the meaning of equality remains vigorous. Famously, Peter Westen argued in an influential article that equality is an ‘empty idea’.Footnote 10 According to Westen, for equality to be operationalized as a legal norm, it must reference external standards of justice.Footnote 11 As a result, he claimed that equality was tautological and that protecting individual liberties alone was sufficient to safeguard it.Footnote 12 However, Westen’s argument has faced significant criticism. For one, his critique presupposes a formal understanding of equality, thereby excluding substantive conceptions.Footnote 13 Moreover, even formal equality can hold independent value, particularly in cases where measures draw distinctions that are plainly arbitrary.Footnote 14
There is ongoing debate in the theoretical literature about the purpose of equality and non-discrimination norms. Many scholars draw a strong connection between equality and human dignity. For instance, Jeremy Waldron contends that equality is rooted in the recognition of every individual’s equal self-worth.Footnote 15 Similarly, Deborah Hellman argues that non-discrimination norms serve to protect individuals from demeaning treatment.Footnote 16 Iyiola Solanke suggests that anti-discrimination laws are designed to guard against stigmaFootnote 17 while Benjamin Eidelson emphasizes their role in addressing the disrespect shown towards the personhood of marginalized individuals.Footnote 18
In her recent book, Sophia Moreau presents a pluralist theory of discrimination, arguing that the harm of discrimination can manifest in three distinct ways: through the unfair subordination of individuals, the infringement of their deliberative freedom or the denial of access to a basic good.Footnote 19 Similarly, Sandra Fredman underscores the importance of substantive equality in her work on discrimination law.Footnote 20 She advances a four-dimensional model of substantive equality, comprising a redistributive dimension aimed at redressing existing disadvantages, a recognition dimension focusing on the promotion of dignity and self-worth, a participative dimension fostering political and social inclusion and a transformative dimension facilitating the accommodation of difference and structural change.Footnote 21 Additionally, several scholars contend that non-discrimination norms are designed to protect groups that face systemic or structural disadvantages, often arising from biases, discriminatory preferences or pervasive negative stereotypes.Footnote 22
This English-language discussion of equality remains largely theoretical. Most approaches focus on an understanding of equality as a non-discrimination guarantee designed to protect disadvantaged groups. When judicial case law is cited, it predominantly draws on examples from common law jurisdictions. Expanding the focus to include legal scholarship outside of common law systems reveals conceptions of equality that extend beyond non-discrimination. In France, a ‘republican’ and ‘universalist’ interpretation of equality – skeptical of an emphasis on disadvantaged groups – has been particularly prominent.Footnote 23 In Germany, equality is often understood in formal terms, rooted in an Aristotelian conception of justice.Footnote 24
However, these accounts usually focus on individual jurisdictions. This article seeks to integrate these diverse understandings of equality into a common analytical framework, thereby enriching the international discussion with perspectives from jurisdictions beyond the common law tradition. It primarily draws on judicial interpretations of equality observed across different legal systems. The article identifies four distinct models of equality, demonstrating that judicial approaches to equality often diverge from the non-discrimination model prevalent in common law contexts. Finally, it examines the normative strengths and weaknesses of these models to assess whether a gold-standard can be identified for the operationalization of equality.
Classifying competing understandings of equality
When courts operationalize equality, they also implicitly engage with the role of judicial review. There are various ways to conceptualize the relationship between the judiciary and the legislative or executive branches. The legitimacy of judicial review remains a subject of intense debate within legal scholarship,Footnote 25 and different legal systems have offered divergent answers regarding the role of courts in the political process. Given that the meaning of equality is contested, its operationalization is not merely a technical exercise. Courts must define the principles of comparability and likeness and determine the extent to which equality encompasses issues such as redistribution and affirmative action. It follows that when courts operationalize equality in different ways, these approaches to equality also reflect distinct understandings of the role of judicial review.
This article identifies four distinct models of equality. These models reflect not only different conceptions of equality but also varying levels of judicial deference to the legislature and the executive. The first model is the equality-as-non-discrimination model that we can find in most common law jurisdictions and that is dominant in the theoretical discussion on equality. Courts adhering to this approach focus on distinctions based on certain suspect characteristics. In such cases, they apply a stricter standard of review, whereas distinctions not involving a suspect characteristic are examined under a more deferential standard. Consequently, applicants in equality cases rarely succeed if the challenged distinction does not concern a suspect characteristic.
A second model, prevalent in many continental European jurisdictions, is the equality-as-reasonableness model. Under this approach, courts apply a relatively strict standard of review even in cases that do not involve a suspect classification. While this does not preclude differentiated standards depending on whether a suspect distinction is at issue, cases without a suspect classification still have a reasonable chance of success. The third model is the positive-equality model, which is even less deferential than the equality-as-reasonableness approach. Under this model, courts derive positive obligations from the principle of equality and impose these obligations on the legislature or the executive. The final model, the deferential model, largely leaves the determination of the meaning of equality to the legislature and the executive and refrains from subjecting political decisions to strict judicial scrutiny.
Before delving into the discussion of the different models, a few caveats are in order. First, the identified models are only approximations. There are often meaningful differences in the equality doctrines of courts that adhere to the same model. For instance, while both the Supreme Courts of Canada and the United States follow a non-discrimination conception of equality, they have conceptualized this idea in significantly distinct ways. Second, the boundaries between the models are not always clearly defined. A court may largely follow the equality-as-reasonableness model, yet still place considerable emphasis on protecting disadvantaged groups from discrimination. Likewise, a court that imposes positive obligations on the legislature may nonetheless adhere to either the equality-as-non-discrimination or the equality-as-reasonableness model when assessing state measures that distinguish between different groups of individuals. Nevertheless, classifying these models remains useful. Models necessarily simplify by emphasizing key characteristics.Footnote 26 We can think of them as analogous to maps: just as a map simplifies complex terrain, a large-scale map offers more detail while a small-scale map provides a broader overview. A map that attempted to depict everything at full scale would be largely impractical.
The equality as non-discrimination approach
The equality-as-non-discrimination approach is among the most prominent frameworks in both theoretical discussions of equality and non-discrimination and judicial practice. This approach primarily targets distinctions drawn on the basis of suspect classifications. When such classifications are involved, courts adhering to this model generally apply a stricter standard of review, giving applicants a reasonable chance of success. Conversely, distinctions not based on suspect classifications are subject to a more deferential standard of review. A well-known example of this approach is the tiered scrutiny framework employed by the U.S. Supreme Court in cases under the Equal Protection Clause of the Fourteenth Amendment to the U.S. Constitution. Under this framework, the Court applies heightened scrutiny – either strict or intermediate – when evaluating distinctions based on specific suspect classifications. For instance, raceFootnote 27 and national originFootnote 28 trigger strict scrutiny, while sexFootnote 29 and extramarital birthFootnote 30 prompt intermediate scrutiny. Although heightened scrutiny does not guarantee that a distinction will be found unconstitutional, challenges are more likely to succeed under this standard.Footnote 31 In contrast, in the absence of a suspect classification, the Court applies the more deferential rational basis test.Footnote 32 According to a common formulation of the test, an equal protection challenge is unsuccessful ‘if there is any reasonably conceivable state of facts that could provide a rational basis for the classification’.Footnote 33 Cases reviewed under the rational basis test rarely succeed.Footnote 34
The equality-as-non-discrimination approach is also widely applied by courts outside the United States, particularly in common law jurisdictions. One example is the equality jurisprudence of the Supreme Court of Canada under Section 15 of the Canadian Charter of Rights and Freedoms.Footnote 35 The Canadian Supreme Court has consistently required that distinctions be based on a suspect classification – that is, on grounds either explicitly listed in Section 15 (1) of the Charter or analogous to them.Footnote 36 No case under Section 15 (1) has succeeded without involving a suspect classification.Footnote 37
Similarly, the U.K. Supreme Court (formerly the House of Lords) follows an equality-as-non-discrimination approach when addressing differential treatment under Article 14 of the ECHR. The Court requires that such treatment be based on ‘status’,Footnote 38 a concept referring to grounds either expressly enumerated in Article 14 or analogous to them. Initially, the U.K. Supreme Court interpreted this term narrowly, restricting it to personal characteristics.Footnote 39 More recently, however, the Court has taken a broader view, extending the concept beyond traditional suspect classifications.Footnote 40 The Court also adjusts the standard of review in its justification analysis depending on the classification involved. For distinctions based on grounds such as sex or sexual orientation, the Court demands ‘very weighty reasons’.Footnote 41 For others, it applies a more deferential ‘manifestly without reasonable foundation’ test.Footnote 42 In practice, cases not involving suspect classifications rarely succeed.Footnote 43
A key advantage of the equality-as-non-discrimination approach is its capacity to address the ‘likeness’ challenge posed by Peter Westen.Footnote 44 Rather than focusing on whether individuals are alike, it emphasizes the inherently problematic nature of certain distinctions. For example, distinctions based on race are considered problematic regardless of whether the individuals being compared are otherwise similar. However, this approach also presents several difficulties, four of which will be discussed in more detail below.
The first challenge concerns the identification of suspect criteria. While the problematic nature of some criteria, such as race, seems evident, other cases are less clear and require abstract principles to define suspect classifications. The second challenge concerns indirect discrimination. If non-discrimination norms address only direct discrimination, they can easily be circumvented by the use of alternative, facially neutral criteria that are strongly correlated with suspect classifications. Yet, the extent to which indirect discrimination is covered remains a matter of ongoing debate. The third challenge arises in relation to affirmative action. If distinctions based on certain characteristics are inherently suspect, can positive measures intended to benefit disadvantaged groups still be justified? Finally, there is the question of whether the equality-as-non-discrimination approach is under-inclusive, as it may fail to address problematic distinctions that do not involve suspect classifications.
Identifying suspect criteria
The equality-as-non-discrimination approach faces its first major challenge in identifying suspect criteria. When a constitution includes a non-discrimination clause in its fundamental rights catalog, it often enumerates specific problematic grounds of distinction. However, such lists are typically open-ended, allowing courts to expand them. Moreover, some courts apply the equality-as-non-discrimination framework even when the relevant constitutional provision is a broader equality guarantee. A prominent example is the Equal Protection Clause of the Fourteenth Amendment, which guarantees ‘the equal protection of the laws’. This raises the question of which criteria of distinction warrant a heightened standard of review.
A comparative examination of case law reveals that judicial reasoning in identifying suspect criteria is often apodictic. One influential reference point is certainly the famous footnote 4 in Carolene Products, which called for ‘more searching judicial inquiry’ when ‘discrete and insular minorities’ are affected.Footnote 45 Yet this statement was merely obiter dictum. In many cases, the reasoning is more restrained. When the U.S. Supreme Court introduced intermediate scrutiny for sex discrimination in Craig v. Boren, Justice Brennan justified the new standard by referring to the Court’s earlier decision in Reed v. Reed. Footnote 46 However, although Reed concerned sex-based distinctions, it was decided under rational basis review.Footnote 47 Similarly, when the European Court of Human Rights (ECtHR) recognized sexual orientation as a ground for discrimination, it simply stated that sexual orientation is ‘a concept which is undoubtedly covered by Article 14 of the Convention’.Footnote 48
To better understand what constitutes a suspect criterion, it is helpful to turn to the theoretical literature. Considerable debate exists over what defines the suspect nature of a distinguishing characteristic.Footnote 49 In her seminal textbook on anti-discrimination law, Sandra Fredman identifies four factors that may determine the suspectness of a classification.Footnote 50 These factors are immutability, access to the political process, dignity and historical disadvantage.Footnote 51 Fredman argues that each of these factors has weaknesses and that no single element can definitely determine whether a criterion of distinction should be regarded as suspect.Footnote 52 Instead, she proposes that courts consider all of these factors and make a context-sensitive determinations in light of the concrete circumstances of each case.Footnote 53
In his influential monograph A Theory of Discrimination Law, Tarun Khaitan advances two cumulative criteria.Footnote 54 First, there must be a significant advantage gap between groups.Footnote 55 Second, the distinguishing characteristic must either be immutable or reflect a fundamental choice.Footnote 56 The first criterion corresponds to two of Fredman’s factors – the lack of access to the political process and historical disadvantage or marginalization.Footnote 57 These considerations are indicative of, and reflect, structural disadvantages that certain groups face within society and the political process.
The second criterion concerns the legitimacy of the disadvantage gap. Without this qualification, even competitive electoral systems could be regarded as discriminating against political ideologies that consistently lose. Distinctions based on immutability have long been considered particularly invidious.Footnote 58 Characteristics such as skin color or sex are difficult or impossible to change, preventing individuals from easily aligning with the majority.Footnote 59 However, immutability alone is too narrow.Footnote 60 Characteristics like religion, while technically changeable, form a core part of an individual’s identity, making compelled change normatively unacceptable. To account for this, Khaitan includes fundamental choices within the scope of non-discrimination protections.Footnote 61
Despite the strength of these criteria, borderline cases arise in which it is unclear whether a characteristic satisfies both conditions. Consider age: it may generate significant advantage gaps depending on context, and it is immutable in the sense that individuals cannot choose their age.Footnote 62 Yet, because most people experience all life stages, age groups are transitory and thus differ fundamentally from characteristics such as race or sex. Comparative judicial practice reflects this ambiguity. Age treated as a suspect characteristic in some jurisdictions, such as the European Union,Footnote 63 but not in others.
Nationality presents another complex example. Non-nationals often face disadvantage gaps due to their lack of political power. Although not entirely immutable, acquiring new citizenship can involve significant legal barriers, and many individuals may choose to retain their birth nationality because of its deep connection to their identity. For this reason, the U.S. Supreme Court has applied strict scrutiny to distinctions based on national origin in certain cases.Footnote 64 However, distinctions based on nationality are commonly accepted in some contexts. Voting rights are usually (though not universallyFootnote 65) limited to nationals. Furthermore, only nationals have an automatic right of residence. This ambiguous status of nationality was noted by Lady Hale in R (Ali) v Secretary of State for the Home Department, where she observed that ‘in the context of immigration, nationality is not a particularly “suspect” classification’.Footnote 66 However, the quote also highlights that there may be other contexts where nationality is a suspect characteristic. This discussion demonstrates that whether a characteristic is considered suspect often depends on context and remains a matter of continuing debate.
Indirect discrimination
Another contentious issue concerns how to address indirect discrimination.Footnote 67 Limiting heightened scrutiny for suspect classifications to cases of explicit discrimination would make it easy to circumvent anti-discrimination protections.Footnote 68 A decision-maker could simply rely on a facially neutral characteristic closely correlated with a suspect classification to target a disadvantaged group. Consequently, it is widely accepted that intentional discrimination is prohibited even when the criterion appears neutral on its face.Footnote 69 However, significant debate remains over the extent to which indirect discrimination is prohibited beyond intentional cases. In the United States, the Supreme Court requires proof of discriminatory intent before applying the heightened scrutiny in constitutional equality cases.Footnote 70
However, discriminatory intent is often difficult to prove.Footnote 71 Moreover, there are strong normative reasons to protect disadvantaged groups from indirect discrimination even in the absence of discriminatory intent.Footnote 72 Accordingly, the restrictive approach of the U.S. Supreme Court has been largely rejected outside of the United States.Footnote 73 Many courts accept that a policy producing a disparate impact on a disadvantaged groupFootnote 74 can amount to indirect discrimination even without evidence of intent.Footnote 75 For example, the Court of Justice of the European Union (CJEU) has famously held that disadvantaging part-time employees compared to full-time employees constitutes indirect discrimination based on sex, given that most part-time employees are women, while men disproportionately occupy full-time positions.Footnote 76
Even if a disparate impact on a disadvantaged group is considered discriminatory in principle, this does not mean every case is straightforward. Legitimate policy choices can sometimes produce a disparate impact, and it is not always obvious that this constitutes indirect discrimination. Consider a government subsidy program for start-ups. If the program prioritizes AI start-ups to promote future technologies, and the AI start-up sector is male-dominated while female entrepreneurs focus more on fields like health or consumer-goods industries, the program would disproportionately benefit men. Would this constitute indirect discrimination against women?
In judicial practice, courts rarely address this issue explicitly. When they do, it is typically analyzed within the general justification framework. A policy that produces a disparate impact on a disadvantaged group may be justified if it pursues a legitimate aim and is proportionate.Footnote 77 However, the proportionality test involves balancing competing normative values, and there are no clear standards for determining how those values and interests should be weighed in a balancing test.Footnote 78 In the case of the start-up subsidy scheme, for example, the test offers little guidance on how to balance the goal of investing in future technologies against the principle of equal opportunities for female entrepreneurs.
To address these challenges, several legal scholars have proposed a more theoretically grounded approach.Footnote 79 They argue that disparate impact becomes problematic when it reflects structural inequalities in society.Footnote 80 Under this framework, the CJEU’s jurisprudence on part-time employeesFootnote 81 is well justified, since women disproportionately work part-time owing to greater childcare responsibilities. This pattern reflects traditional gender roles and, therefore, structural inequalities.
Assessing the start-up subsidy scheme, however, is more complex. One might argue that male dominance in the AI sector reflects gender-based differences in preferences and interests. Yet this explanation oversimplifies the issue. On the one hand, AI might be seen as more deserving of subsidies because political decision-making, like the AI sector, is male-dominated. This could mean that male political dominance shapes policy preferences favoring male-dominated industries. On the other hand, preferences are not inherently male or female but are socially constructed and shaped by cultural expectations. These latter considerations suggest that the subsidy scheme might also reflect structural discrimination. However, the example also underscores the difficulty of applying the concept of structural discrimination in practice – even if it is compelling in theory.
Affirmative action and positive measures to support disadvantaged groups
A third challenge concerns the treatment of positive measures: To what extent is it permissible to draw distinctions based on a suspect classification to benefit disadvantaged groups? This issue has been particularly contentious in the United States, where the Supreme Court applies strict scrutiny to affirmative action policies.Footnote 82 In Regents of the University of California v Bakke, Justice Powell asserted that ‘[r]acial and ethnic distinctions of any sort are inherently suspect and thus call for the most exacting judicial examination’.Footnote 83 Although this does not mean that all affirmative action programs for racial minorities automatically violate the Equal Protection Clause, the high bar set by strict scrutiny is often difficult to meet.Footnote 84
Most jurisdictions outside the United States take a more accommodating stance towards positive measures. For instance, Section 15 (2) of the Canadian Charter was drafted partly in response to the U.S. Supreme Court’s restrictive position on affirmative action.Footnote 85 However, even in jurisdictions more receptive to such measures, the question of their permissible scope remains contested. Conceptually, there are two principal ways to prevent positive measures from being classified as violations of equality. One approach applies an asymmetric conception of discrimination that excludes positive measures from the definition of discrimination altogether. Alternatively, they can be considered prima facie discriminatory but ultimately justified under a justification analysis.
The first approach emphasizes the purpose of non-discrimination norms and advocates for their asymmetric application.Footnote 86 If these norms are designed to protect disadvantaged groups, it follows that individuals outside those groups should not be able to claim discrimination. However, this reasoning has its limitations. Disadvantage is context-dependent and relative: a group may be more disadvantaged than one but less disadvantaged than another. Consider, for example, Asian Americans seeking admission to elite U.S. universities. Arguably, they are socially more disadvantaged than white Americans but less disadvantaged than African Americans. Second, measures that formally disadvantage members of a non-disadvantaged group can still perpetuate harmful stereotypes. Consider, for example, policies granting automatic survivor’s benefits to the widows of deceased male employees but denying the same benefits to widowers of deceased female employees.Footnote 87 Such rules technically disadvantage men, yet they also reflect and reinforce the problematic stereotype of men as the primary breadwinners.
The second approach focuses on the justification analysis.Footnote 88 Under this framework, positive measures benefiting disadvantaged groups are considered prima facie discriminatory but may be justified. While this approach offers greater flexibility, it also poses challenges. Notably, courts must develop clear standards for determining which types of positive measures are permissible. The debates surrounding women’s quotas and preferential university admissions illustrate the complexity of assessing positive measures on a case-by-case basis, even when they are deemed legitimate in principle. As a result, many courts – even outside the United States – frequently conclude that certain positive measures violate non-discrimination norms. A notable example is the jurisprudence of the CJEU, which has imposed limits on the use of women’s quotas in public-sector hiring.Footnote 89
Under-inclusiveness
The final challenge of the non-discrimination model lies in its potential under-inclusiveness. Some scholars contend that focusing exclusively on suspect classifications risks excluding other problematic distinctions from constitutional equality guarantees.Footnote 90 Adopting a less deferential standard of review for cases not involving a suspect classification would shift the framework towards the equality-as-reasonableness model, which is discussed below. However, this shift would also introduce the normative issues associated with that approach, particularly the lack of clear standards for justification analysis. Nevertheless, most jurisdictions do not entirely dismiss non-suspect equality claims. Courts adhering to the equality-as-non-discrimination model often address instances of arbitrary discrimination. For instance, many successful rational basis cases before the U.S. Supreme Court arguably address situations of arbitrary administrative or legislative actions.Footnote 91 Similarly, international human rights treaty bodies, which generally adhere to a non-discrimination approach, frequently correct arbitrary state conduct. Yet, the concept of arbitrariness remains elusive. It refers to decisions whose unreasonableness is glaringly obvious. However, lawyers can reasonably disagree on where to draw the line between truly arbitrary measures and those that are merely poor policy choices.Footnote 92
The equality as reasonableness framework
The second major framework for understanding equality is the equality-as-reasonableness approach. Under this framework, any differentiation between comparable cases – regardless of the specific differentiating factor – raises a prima facie equality concern and places a significant burden of justification on the government or legislature.Footnote 93 Equality doctrines based on the equality-as-reasonableness framework typically build on the Aristotelian principle that like cases should be treated alike and different cases differently.Footnote 94 Once a court identifies unequal treatment of similar cases or equal treatment of dissimilar ones, it requires the government or the legislature to provide a justification for the distinction. This framework typically leads to a relatively high number of successful equality claims, even in cases that do not involve suspect classifications. For example, it applies to distinctions between employees and public servantsFootnote 95 or between pharmacies located in train stations and other retail establishments within the same premises.Footnote 96
Many judicial decisions applying the equality-as-reasonableness standard frequently review the rationality of tax and social-benefits legislation. For example, the German Federal Constitutional Court held in its ‘pension taxation’ decision that taxing public servants’ pensions differently from those of employees under the general pension insurance scheme violated the equality clause of the German Constitution.Footnote 97 Similarly, the Polish Constitutional Tribunal extended the scope of a special pension scheme for police officers and other uniformed officials.Footnote 98 It held that customs officers were comparable to police officers, and excluding them from the scheme violated the constitutional guarantee of equality.Footnote 99
These examples illustrate that the equality-as-reasonableness framework is less deferential than the equality-as-non-discrimination approach when it comes to cases not involving suspect criteria. In particular, the level of review is more demanding than the rational basis test in the tiered scrutiny model of the U.S. Supreme Court. As the following discussion will show, many of the challenges associated with the equality-as-non-discrimination approach do not apply as strongly to the equality-as-reasonableness model, as it places less emphasis on suspect classifications. However, it raises different challenges – most notably its susceptibility to Peter Westen’s critique of equality as an ‘empty formula’.Footnote 100
The role of suspect classifications
Even though courts applying the equality-as-reasonableness approach generally show less deference towards distinctions not based on suspect classifications than courts following the equality-as-non-discrimination approach, they do not necessarily employ a uniform standard of review. Instead, they often apply a stricter standard of review to distinctions based on suspect characteristics. For example, the German Federal Constitutional Court uses a sliding scale (‘tierless’) formula to operationalize the general constitutional equality guarantee. According to this formula, it applies a ‘continuous constitutional standard of review based on the principle of proportionality’.Footnote 101 In practice, the Court mainly distinguishes between two levels of review.Footnote 102 Depending on the concrete distinction, it applies either a more ‘generous standard of review’Footnote 103 or ‘a stricter standard of review’Footnote 104.
The main difference to the equality-as-non-discrimination standard lies in the relative importance of non-suspect distinctions. While successful cases that do not involve a distinction based on a suspect characteristic are rare under the equality-as-non-discrimination model, they constitute the majority of successful claims under the equality-as-reasonableness model. For examples, although the German Federal Constitutional Court has issued some important decisions concerning discrimination based on sex or sexual orientation,Footnote 105 most successful equality challenges to legislation concern distinctions unrelated to a suspect classification – particularly in the areas of tax and social benefits.Footnote 106 Similarly, in Poland and France, the vast majority of distinctions found to violate the constitutional equality guarantee are not based on suspect classifications.Footnote 107 Consequently, if the success of a case does not depend strongly on whether the differentiating characteristic is qualified as suspect or not, the precise definition of what constitutes a suspect characteristic becomes less relevant.
Indirect discrimination
Because the identification of a suspect criterion is less central under the equality-as-reasonableness approach than under the equality-as-non-discrimination model, the concept of indirect discrimination has long received little attention. More recently however, courts have recognized that it can nevertheless serve as a useful analytical framework. The German Federal Constitutional Court introduced the concept in a 1997 decision concerning distinctions between full-time and part-time employees in the context of old-age pension benefits for public servants.Footnote 108 The Court held that a distinction constitutes indirect discrimination ‘based on gender even if the regulation is formulated in a gender-neutral manner, but primarily affects women’.Footnote 109 Similarly, the French Constitutional Council recognized the concept of indirect discrimination in the early 2000s.Footnote 110 By contrast, the Polish Constitutional Tribunal has, to date, not developed a specific doctrine addressing cases of indirect discrimination.
Affirmative action and positive measures to support disadvantaged groups
Conceptually, positive measures to support disadvantaged groups should not pose significant difficulties under the equality-as-reasonableness approach, as such questions can be analyzed within the general justification framework. However, there is no unified approach to positive measures among jurisdictions following this approach. The Polish Constitutional Tribunal has explicitly accepted positive measures for the support of women as ‘compensatory privileges’ aimed at reducing gender inequality.Footnote 111 In Germany, positive measures to achieve gender equality are recognized by Art. 3, para. 2 of the German Constitution. Nevertheless, the Federal Constitutional Court has rarely had the opportunity to rule directly on the issue.Footnote 112 By contrast, the French Constitutional Council has, on several occasions, invalidated gender quotas enacted by the legislature and required constitutional amendments to justify them.Footnote 113 This jurisprudence reflects a formal and symmetrical understanding of equality often associated with the so-called ‘French republican model’, which is widely perceived as hostile to the recognition of legal group rights.Footnote 114
Under-inclusiveness
The principal advantage of the equality-as-reasonableness model is that it avoids the problem of under-inclusiveness identified in relation to the equality-as-non-discrimination model. Its less deferential stance towards distinctions that do not involve suspect classifications ensures that all potentially problematic differentiations are subject to scrutiny, extending beyond merely correcting arbitrary measures. However, the standard for assessing distinctions that are not based on suspect classifications remains largely indeterminate. Once a court determines that comparable cases have been treated differently – or that dissimilar cases have been treated alike – courts impose a burden of justification on the government or the legislature. As Westen has demonstrated, the key difficulty lies in determining what constitutes likeness.Footnote 115 No two individuals or situations are identical in all respects.Footnote 116 The task, therefore, is to identify the relevant characteristics for comparison. Yet the selection of these comparator characteristics entails a normative judgment external to the concept of equality itself.Footnote 117
At first glance, a potential solution to this problem is a broad interpretation of likeness. For instance, the German Federal Constitutional Court typically applies an expansive notion of likeness, ensuring that equality challenges rarely fail at this stage.Footnote 118 Instead, the Court shifts the focus to the justification analysis, determining whether the differential treatment is warranted. However, this approach does not resolve the problem – it merely relocates it from the determination of likeness to the justification analysis. The Court’s reasoning illustrates this: it holds that unequal treatment can be justified if ‘there are […] differences between the two groups of such a nature and such a weight that they can justify different treatment’.Footnote 119 This formulation highlights the core problem – it remains unclear how the Court establishes the normative standard for determining whether the differences between the groups are sufficiently significant to justify unequal treatment.
The German Federal Constitutional Court’s ruling in the ‘pension taxation’ case serves as an apt illustration of this point.Footnote 120 The case addressed differences in the taxation of pensions for public servants versus those under the general pension insurance scheme. In Germany, most employees are enrolled in the general pension insurance system and contribute income-dependent amounts. Public servants, however, do not pay into that system; their pensions are instead financed through tax revenues. As a result, pensions from the general insurance scheme were partially tax-exempt, while public servants’ pensions were fully taxed. The Federal Constitutional Court ruled that this disparity in taxation was unjustified, finding a violation of the constitutional equality guarantee.Footnote 121
The Court emphasized inconsistencies in the pension system, particularly noting that the general pension insurance scheme was heavily subsidized by tax funds.Footnote 122 Nevertheless, it is remarkable that the Court – while repeatedly acknowledging the difficulty of comparing the two schemes –Footnote 123 still found that they were sufficiently similar to find that the unequal treatment was unjustified. This case demonstrates that the flexibility of the equality-as-reasonableness framework not only grants courts substantial discretion but also opens the door to potential judicial overreach. As a result, courts employing this framework assert broader powers of judicial review than those limiting themselves to cases involving suspect classifications.
The positive-equality framework
The third approach is the positive-equality framework. Unlike the previous two, it goes beyond addressing state measures that draw distinctions. It also imposes positive obligations on states to actively protect disadvantaged groups. In this respect, it shares similarities with the equality-as-non-discrimination model, in which affirmative action is a subject of considerable debate. However, while the affirmative action debate focuses on the extent to which legislatures or other authorities may adopt positive measures favoring disadvantaged groups, the positive-equality model grants courts the authority to mandate the government or legislature to implement such protective measures. These obligations can take various forms – for example, requiring the state to protect women against gender-based violence,Footnote 124 ensure workplace safety for vulnerable workers,Footnote 125 extend existing protections to other disadvantaged groups, or even establish social programs to support individuals from marginalized backgrounds.
This approach is predominantly found in several Latin American jurisdictions, such as Brazil or Colombia, as well as in the jurisprudence of the Inter-American Court of Human Rights (IACtHR).Footnote 126 A paradigmatic example is the Fireworks Factory case decided by the IACtHR.Footnote 127 The case concerned an explosion at a fireworks factory that killed more than sixty workers, most of whom belonged to vulnerable groups – predominantly poor, Black and female individuals. The Court held that Brazil had violated the equality norm under Article 24 of the ACHR. It ruled that this provision guarantees material equality and requires ‘the adoption of positive measures of promotion in favor of groups that have historically been discriminated against or marginalized’.Footnote 128 The state had failed to ‘reverse the situation of structural poverty and marginalization of the fireworks factory workers’Footnote 129 and to ‘ensure that real measures were taken to protect the life and health of the workers and to guarantee their right to material equality’.Footnote 130
Another illustrative example is ADO 26, a case before the Brazilian Supreme Court (Supremo Tribunal Federal – STF).Footnote 131 In that case, the STF derived a legislative obligation from the constitutional equality clause, requiring the enactment of criminal law provisions to protect victims of homophobia and transphobia.Footnote 132 Not every court that derives positive obligations from equality necessarily falls into the positive-equality category. For example, the ECtHR has, on several occasions, derived duties to protectFootnote 133 and to accommodateFootnote 134 disadvantaged individuals from Art. 14 ECHR. However, these positive obligations are relatively concrete, narrowly defined and far more limited in scope than the positive obligations that were developed by the IACtHR in the Fireworks Factory case.
The role of suspect criteria
Because the positive-equality model places a strong emphasis on disadvantaged groups, the identification of suspect criteria plays an important role. For example, the IACtHR applies an enhanced level of scrutiny – and even a presumption of discriminatory treatment – if a differentiation is based on a suspect classification.Footnote 135 The Colombian Constitutional Court also applies different levels of scrutiny.Footnote 136 It employs the strictest level of scrutiny to distinctions based on a suspect classification or affecting groups that are marginalized, discriminated against or manifestly vulnerable.Footnote 137 By contrast, the Brazilian Supreme Court applies an equality test that does not formally distinguish between different tiers of review. Nonetheless, it has recognized certain groups as deserving special protection due to historical marginalization.Footnote 138 These groups may not be targeted through differentiation based on suspect grounds and may even benefit from positive measures that the court derives from the constitutional equality guarantee.Footnote 139
Positive equality
As already noted, courts adhering to the positive-equality model are highly accommodating towards positive measures. On the one hand, they typically accept distinctions based on suspect classifications as justified if such distinctions aim to support disadvantaged groups.Footnote 140 However, what is specific for courts following the positive-equality framework is that they go beyond merely accepting positive measures enacted by the legislature. In some cases, they extend the benefits of such measures to additional disadvantaged groups.Footnote 141 In other cases, they derive independent positive obligations directly from constitutional equality guarantees.Footnote 142
Strengths and weakness of the model
In several respects, the positive-equality framework is even less deferential than the equality-as-reasonableness model. As David Landau has observed, courts in the Global South often adopt a more interventionist posture to compensate for the deficiencies of dysfunctional political institutions.Footnote 143 Structural inequalities are particularly pronounced in Latin America,Footnote 144 where the capacity and effectiveness of democratic legislatures are frequently limited.Footnote 145 Consequently, many Latin American courts perceive themselves as transformational social actors, playing an active role in redressing these disparities.Footnote 146 It is therefore unsurprising that courts in the region have embraced a distinctly non-deferential approach.Footnote 147
However, this broad conception of judicial power entails significant drawbacks, as courts risk assuming functions traditionally reserved for the legislature. In the academic discussion on European integration, some scholars have proposed a useful distinction between regulative and redistributive measures.Footnote 148 They argue that regulatory functions can be effectively strengthened by delegating authority to independent agencies or courts.Footnote 149 In contrast, redistributive policies require a higher degree of democratic legitimacy and should therefore remain within the domain of democratic decision-making.Footnote 150 While positive measures derived from equality do not always fall squarely within the redistributive category, they often do. When courts mandate the expansion of social programs or the allocation of financial resources for disadvantaged groups, their decisions can have substantial redistributive consequences. This is not inherently illegitimate, particularly in contexts where democratic institutions are severely impaired. However, such interventions impose a heavy burden of justification on the courts.
Moreover, there are multiple strategies for addressing structural discrimination, each involving its own set of trade-offs. Reasonable disagreement may arise, for instance, over whether race- or gender-based quotas constitute appropriate tools to redress discrimination and compensate for historical injustices. Such affirmative action measures can inadvertently reinforce the salience of the very categories they seek to neutralize, potentially entrenching their significance. Additionally, while these positive measures are designed to promote inclusivity, they can also produce exclusionary effects by prioritizing certain disadvantaged groups while overlooking others who, though less disadvantaged, still experience marked inequalities compared to the majority population. It is therefore questionable whether courts are the most suitable institutions to navigate these complex trade-offs. As a result, the legitimacy of a positive-equality approach is highly context-dependent. In systems with well-functioning legislatures, there is little need for an interventionist judiciary. Conversely, in political systems characterized by significant legislative or executive dysfunction, a more assertive role of courts in advancing equality may be necessary and justifiable.Footnote 151
The deferential approach
The final framework is the deferential approach, which accords significant deference to the legislature in all equality cases. Under this model, courts rarely find a measure incompatible with constitutional equality guarantees. A key example is the jurisprudence of the Singapore Supreme Court, comprising the High Court and the Court of Appeal. Traditionally, the Court of Appeal has emphasized a general presumption of constitutionality.Footnote 152 As a result, applicants claiming a violation of the equality clause in Article 12 of the Singapore Constitution must first establish a prima facie case of discriminatory intent.Footnote 153
The justification for the deferential approach lies in a particular understanding of the separation of powers. When the meaning of equality is contested, it is seen as the legislature’s – not the judiciary’s – role to define its substantive content. This principle was articulated in the Lim Meng Suang case, a case concerning the criminalization of sexual intercourse between homosexual men.Footnote 154 The Court held that it lacked the authority to second-guess the legislature’s rationale for such differentiation, emphasizing its reluctance to engage in judicial activism.Footnote 155 It reasoned that it was the legislature’s responsibility ‘to ensure that the Singapore Constitution reflects the prevailing social mores as well as the aspirations of Singapore society’.Footnote 156 This reflects the non-confrontational approach of Singapore’s judiciary: in several lectures, successive Chief Justices have emphasized that the country’s judicial culture prioritizes avoiding conflicts with the political branches.Footnote 157
More recently, the Court has somewhat relaxed its previously rigid deference to the political branches. In Syed Suhail, it adopted a more rigorous standard of review for cases that ‘affect the appellant’s life and liberty to the gravest degree’.Footnote 158 Nevertheless, the stringent standards imposed by Singaporean courts continue to place a heavy burden on applicants – one that is rarely overcome in practice. This is reflected in the absence of a single successful equality claim in Singapore’s jurisprudence.
The role of suspect criteria
Even if courts apply a deferential standard of review across the board, the presence of suspect criteria of differentiation can, in theory, still make a difference. It is conceivable that a court applies a very deferential standard when a differentiation does not involve a suspect classification, yet remains prepared to correct particularly egregious suspect distinctions. In practice, however, the Singapore Supreme Court has had very few cases involving suspect classifications. Most prominently, the Court had to deal with a discrimination on the basis of sexual orientation in two cases that were directed against a criminal statute penalizing sexual relations between homosexual men.Footnote 159 In the already mentioned Lim Meng Suang case, the Court adopted a narrow interpretation of suspect classifications. It reasoned that sexual orientation did not fall under this category because it was not explicitly mentioned in the non-discrimination clause of Art. 12 (2) of the Singapore Constitution.Footnote 160
Indirect discrimination
Under a deferential approach, courts can generally be expected to show reluctance in recognizing indirect discrimination claims involving a disparate impact on disadvantaged groups. This expectation was confirmed in the second Syed Suhail case, decided by the High Court of Singapore.Footnote 161 The applicants, who were citizens of Malay ethnicity, had argued that their death sentences for drug offenses were discriminatory, as Malays were vastly overrepresented among those convicted of drug-related crimes compared to their share in Singapore’s general population.Footnote 162 However, the Court required a ‘causal link’ between the Advocate General’s prosecutorial decisions and their ethnicity, finding that the statistical evidence provided by the applicants did not establish such a connection.Footnote 163
Positive measures
While the Singaporean courts are reluctant to recognize claims of indirect discrimination, they are generally accommodating towards positive measures enacted by the legislature or executive. In Ravi s/o Madasamy, the Court held that a constitutional amendment designed to increase the likelihood of minority candidates being elected in presidential elections did not violate the non-discrimination clause of Art. 12 (2) of the Constitution because it was justified as a means of ensuring the representativeness of the presidency in a ‘multiracial society’.Footnote 164
Informal signalling
The deferential approach of the Singapore Supreme Court does not mean that the courts are entirely powerless when confronting the political branches. Instead, they occasionally employ informal means to signal inconsistencies between government policies and the constitution.Footnote 165 This dynamic is illustrated by the recent Tan Seng Kee decision.Footnote 166 Like Lim Meng Suang, this case also concerned the criminalization of sexual relations between homosexual men. However, this time, the Court adopted a more nuanced stance. Rather than ruling on the constitutionality of the provision in the Singapore Criminal Code, it held the law effectively unenforceable, reasoning that the applicants had legitimate expectations that it would not be applied.Footnote 167 As a result, the applicants were found to lack standing.Footnote 168 In response to this decision, the Singapore legislature repealed the challenged provision – demonstrating its attentiveness to the subtle signals sent by the Court of Appeal.
The deferential approach has significant shortcomings, particularly regarding the effectiveness of judicial review. By placing trust in the political branches to define equality impartially, it presumes that they will do so without bias – an assumption that is not always warranted, even in liberal democracies. Political decisions can be influenced by implicit bias, stereotypes or entrenchment of structural discrimination, especially where disadvantaged minorities lack meaningful representation. In such cases, courts serve an essential corrective function, addressing the ‘market failures’ of the political process.Footnote 169 Yet an overly deferential approach inhibits them from fulfilling this role as guardians of constitutional equality.
Conclusion
The preceding analysis highlights the remarkable diversity in how courts approach constitutional guarantees of equality. Some courts adopt a deferential stance, while others apply a moderate level of deference, focusing primarily on discrimination based on suspect grounds. Some courts take a less deferential approach, employing an equality-as-reasonableness framework to assess the rationality of legislative or executive decisions. The most expansive model embraces a positive understanding of equality, asserting that constitutional equality norms impose positive obligations on the state to protect disadvantaged groups.
This diversity of approaches challenges the influential narrative of a global model of constitutional rightsFootnote 170 – at least for the field of equality. It raises the question of why we do not see a greater convergence in doctrinal approaches. The answer likely lies in the differing normative strengths and weaknesses of the various models. None of the frameworks discussed holds an absolute normative advantage over the others. Rather, their relative merits depend on the cultural and institutional context in which a court operates.
Take, for example, the deferential approach. This model may function effectively in Singapore, but it relies on well-functioning institutions, high levels of social trust and a culture of inter-institutional respect. Therefore, it is difficult to see how this model could serve as a universal template for other jurisdictions. At the opposite end of the spectrum, the positive-equality framework emerged in contexts where a long histories of military dictatorship or government repression have fostered a deep scepticism of government authority. In such environments, courts were often tasked with serving as a bulwark against authoritarianism, leading to a transformative self-understanding in which they seek to compensate for chronically weak political institutions.Footnote 171
The equality-as-non-discrimination framework is most commonly found in common-law jurisdictions. Given their historical trajectories and self-identification as immigrant societies, these jurisdictions tend to be more diverse and pluralistic.Footnote 172 In such contexts, the non-discrimination approach provides a framework for managing relationships among different social groups. By contrast, the equality-as-reasonableness framework predominates in continental Europe. Many European jurisdictions maintain a strong faith in judicial expertise and uphold a firm conceptual separation between law and politics.Footnote 173 This confidence in judicial rationality supports a less deferential approach, reflected in the equality-as-reasonableness model.
Consequently, it is difficult to establish a clear normative hierarchy among the various models of equality. The suitability of each framework depends on the cultural and institutional context of the jurisdiction in which it operates. This does not mean that courts never get it wrong. On the contrary, rigorous scholarly engagement with judicial equality jurisprudence remains essential – even if we acknowledge that there is no single, universally correct understanding of equality. Any critique, however, must take into account the specific context and institutional environment in which courts develop their doctrine.
Acknowledgements
The article results from a project that has received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (grant agreement No 817652) as part of the ERC Consolidator Grant ‘Correcting Inequality through Law’. I am deeply grateful to Vicente Benitez-Rojas, Faisal Bhaba, Jacco Bomhoff, Gráinne de Búrca, Aparna Chandra, Konstantin Chatziathanasiou, Sujit Choudhry, Erin Delaney, James Fowkes, Carlos Gaio, Michaela Hailbronner, Samuel Issacharoff, Neha Jain, Tarunabh Khaitan, Malcolm Langford, Sonia Lawrence, Nora Markard, Jud Mathews, Janet McLean, Sophia Moreau, Jaclyn Neo, Arie Rosen, Anna Śledzińska-Simon, Richard Stacey, Emanuel Towfigh, Joseph Weiler and Gentian Zyberi for helpful comments, suggestions, critique on the project in various stages of its development. Furthermore, I am grateful to the assistance of Paula Aguilar Sievers, Jasmin Beck. Olivia Bernhörster, Ingo Borgdorf, Parya Darafshan, Paul Leonard Enderle, Johanna Fischer, Felix Fouchard, Eileen Friederichs, Tainá Garcia Maia, Julia Gosing, Eva Marie Janke, Helena Kaim, Dilan Kilic, Antonia Lawrenz, Luca Katharina Lichter, Annkathrin Lindert, Nele Meier, Joanna Mrozinska, Joshua Okeke, Shubhangi Roy, Konrad Schilling, Theo Schütz, Paula Schrage, Sophie Schwab and Helina Teka who supported the project as collaborators and research assistants through the evaluation and coding of cases in various jurisdictions.
Open access
