The Poverty Discrimination Puzzle

I. THE ANALOGY BETWEEN PARADIGMATIC AND SOCIOECONOMIC DISCRIMINATION

In this section we support the analogy (i.e., (P2)). We argue that four prominent rationales that have been given in the literature for prohibiting direct paradigmatic discrimination apply with equal force to direct socioeconomic discrimination.

Consider first ideas of social subordination. Following Sophia Moreau, we understand subordination as a “state of affairs in which one social group has a standing in society as a whole that is lower than that of another social group”.⁴ Discrimination contributes to such subordination. One way in which it does so is through the role of invidious stereotypes and social stigma. Gender inequality is sustained through gender stereotypes which serve to rationalize patterns of inequality. Caste operates through the social stigma associated with ideas of purity and pollution.

Socioeconomic class also operates within a framework of status differences. The poor are not just materially disadvantaged, they also face invidious stereotypes and social stigma. In this respect there are telling parallels between socioeconomic class and race. In the United States the stereotype of being “lazy and undeserving” is associated with both groups. A report from the Center of American Progress states that “it’s notable that labels suggesting laziness or lack of effort that have been used to describe African Americans are also applied to poor people more generally”.⁵ Negative stereotypes of the poor are also manifested through psychological mechanisms of the same kind as those involved in cases of paradigmatic discrimination. One example is stereotype threat. Making salient one’s identity, and the negative stereotypes associated with it, leads one to underperform. Stereotype threat is prominent in connection with sex, race, and caste, but the same mechanism is involved in the case of socioeconomic class.⁶

Lastly, status differences can be gauged from class-related slurs as a means to show disdain. Such slurs – and their prevalence – are akin to sexist or racist slurs. Some, like “welfare queen” and “white trash” in the United States, even have a meaning which intersects between poverty and sex, and poverty and race. Others, like “chav” and “pleb” in the United Kingdom, are class-focused but convey similarly derogatory connotations. Taken together, the discussion establishes that the poor are a subordinated group. Insofar as discrimination law exists to tackle subordination it should cover this form of subordination, too.⁷

The relevance of stigma and stereotypes to ideas of subordination explains why we focus on the more complex idea of “socioeconomic” class, as opposed to a notion of class defined in purely economic terms. This focus fits nicely with sociological work defining class as something composed of economic as well as social and cultural aspects. Social capital is a measure of advantage resulting from one’s social ties. Cultural aspects (specifically cultural capital) include differences and advantages in one’s cultural background and tastes. This broader perspective is important because these cultural and social aspects of socioeconomic class are central to socioeconomic discrimination. Stereotypes and social stigma often attach to the social and cultural avatars of class – accents, cultural preferences, tastes, social connections, and so on.

The shift to socioeconomic class has a second advantage. Purely economic criteria, such as income levels, focus only on one’s current socioeconomic status. We believe that discrimination based on one’s present socioeconomic status is an important form of socioeconomic discrimination. But it is not the only form of socioeconomic discrimination. Socioeconomic discrimination can, and often does, take place on the basis of one’s socioeconomic origin. For example, in Social Class in the 21^st Century, sociologist Mike Savage and colleagues find evidence for a “social class background pay gap” in the United Kingdom.⁸ Savage and colleagues use class as an origin concept, focusing on the professional background of the worker’s parents. As an origin concept, class is something individuals take with them even as they climb the social ladder and retain for life. One retains social contacts, an origin-related identity and some markers, like a distinctive accent.

We now move on to the second of the four rationales for prohibiting direct paradigmatic discrimination. According to Deborah Hellman discrimination is wrong because it demeans, and the further suggestion that such discrimination should therefore be prohibited.⁹ For an act of discrimination to demean a discriminatee, two conditions must be met. First, the action has the expressive meaning that the discriminatee has a lower moral status than others. When Black people are ordered to sit at the back of the bus, then, given the history of Jim Crow, this expresses not simply a seating order but a status hierarchy among passengers. On Hellman’s account, whether an act is demeaning depends on how the act would be understood by a well-placed interpreter of the action’s cultural meaning. The expressive meaning is an objective social fact independent from the discriminator’s intentions, or the discriminator’s own understanding of their action.

The second condition is that the discriminator must be in a position of power relative to the discriminatee. This condition ensures that the discriminatee is not just disrespected by the act which expresses that they have a lower moral status, but that they are put down. Hence, ordinarily while someone at the bottom the hierarchy can disrespect someone at the top, they cannot demean them in Hellman’s sense, because they lack power over them.

How does Hellman’s account differ from Moreau’s? Both accounts link facts about social status and individual actions. For Moreau’s account it matters how individual acts causally contribute to social subordination. For Hellman’s account facts of social status determine the meaning of an individual act. That way, the subordination account lays the emphasis on the state of affair of subordination, while the demeaning account emphasizes individual acts. Second, while there might be an important expressive dimension to subordination, subordination – unlike demeaning – is not all about expression, e.g., it is also about the systemic disregard for certain people’s interest and systemic above-average concern for that of others. Finally, the power aspect in Hellman’s account of what it is to demean is absent in Moreau’s theory of subordination – even a powerless person can causally contribute to reproducing a state of subordination of people like themselves.

Demeaning in Hellman’s sense also occurs in many instances of socioeconomic discrimination. Take the hiring practice, adopted by some firms, of asking applicants about their social background and filtering applications on that basis. In interviews conducted with graduates from some of India’s elite universities, Ashwini Deshpande and Katherine Newman found that interviewers placed great value on the family background of applicants, screening for cues of both caste and socioeconomic class.¹⁰ Applicants sensed that interviewers were keen on a middle- or upper-class backgrounds and looked unfavorably on applicants from poorer backgrounds. Considering the background of social stigma narrated earlier, this practice demeans those who are already socially disadvantaged by expressing the judgment that an applicant’s background is not “respectable” and that they do not belong in a professional setting despite being qualified.

Unlike the two theories examined so far, some theories of discrimination law focus on freedom rather than equality. One example is Tarunabh Khaitan’s theory – the third account we consider. The pervasive disadvantage faced by groups who are subordinated also impacts their freedom to lead a good life. Patterns of discrimination limit negative freedom, opportunities, and access to self-respect – all needed to lead a good life. Prohibiting discrimination by law, then, breaks these patterns and thereby provides members of protected groups, such as women or racial minorities, with access to these basic goods. This consequently enhances their freedom.¹¹

This reasoning is even better suited to explain that to which it is extended (socioeconomic discrimination) than its original object (paradigmatic discrimination). In his account of the point of discrimination law, Khaitan emphasizes that deficiencies in the basic goods necessary for living a good life – on his account, these are negative freedom, an adequate range of valuable opportunities and self-respect – are interconnected in the case of protected groups, so that a deficiency in terms of one results in deficiencies in the other two as well. But if this is true of paradigmatically protected groups, it is also true of the poor. The poor face pervasive disadvantage across multiple dimensions. We should care about the poor not merely because economic deprivation, but because disadvantage clusters. Economic deprivation is linked to deprivation in health, access to social goods, social standing, self-respect, and so on.¹² Taken together these various deprivations lead to the absence of an adequate range of valuable opportunities. Given this multidimensional disadvantage, the freedom of the poor, too, ought to be guaranteed by means of law.

Fourth, and lastly, consider a somewhat different rationale for prohibition relating to freedom: that race, caste, and sex discrimination should be prohibited because they make one’s race, caste, and sex a cost that one needs to consider in deliberating on how to lead one’s life. Sophia Moreau has argued that this infringes our deliberative freedom to lead a life free from such costs.¹³ For members of disadvantaged groups, making certain choices comes with costs for them that are not faced by members of advantaged groups when the latter make the similar choices. Members of disadvantaged groups need to weigh up the reactions of others to their identity. For example, the freedom to be casteless is an exclusive freedom enjoyed by those who are privileged in the caste system.¹⁴ Direct discrimination straightforwardly deprives individuals of this freedom.

Again, something similar can be claimed about social class. Considering the pervasive sense of adverse opinions about the poor which we sketched earlier, the poor need to weigh up people’s judgments about their respectability, and how they fit in, when deciding what to do. Attending university, for a working class, first-generation student, involves navigating other people’s reactions to one’s background. One’s socioeconomic class is a cost here that working class students need to factor in when deciding whether to attend higher education. University environments often value specific cultural forms of expression, including accents, which belong to privileged socioeconomic classes.¹⁵

We have considered four rationales for the prohibition of discrimination from the literature.¹⁶ The first two related to ideas of subordination and inequality; the remaining two connected with ideas of freedom. In all cases, we argued, the reasoning was applicable to socioeconomic discrimination as well. The prohibition on socioeconomic discrimination is therefore supported by a wide range of views on the rationale for discrimination law. This supports the analogy.

II. SOCIOECONOMIC DISCRIMINATION, THE MARKET ECONOMY, AND THE WELFARE STATE

Some might accept the arguments in the previous section and nevertheless hold that it is warranted to outlaw paradigmatic but not socioeconomic discrimination on the ground that the consequences of prohibiting direct socioeconomic discrimination are worse than those for paradigmatic traits. We focus on two reasons. One concerns the impact on the market economy, the other the impact on the welfare state.

Consider a case in which a poor person is denied a good because of their inability to pay. Would this be covered by the prohibition of socioeconomic discrimination? After all, it is because a person is poor that they are unable to pay for the good. In a market economy, a law prohibiting socioeconomic discrimination would therefore appear to clash with the market mechanism. But charging a fair market price should not be rendered illegal. Rather it seems that insofar as we are concerned about some persons being unable to afford as much as others, this is a concern about the distribution of money, not a concern about discrimination.¹⁷

We have two responses. First, we stress that the concerns we have raised about socioeconomic discrimination are not directly about the inability to afford goods. They are about how people from “lower” socioeconomic classes are treated in society. They are subjected to stigma and stereotypes, and suffer from lower social standing, in addition to their material deprivations. Even though the two tend to coincide, it is possible for a society to be materially unequal but free of such status differences. Jiwei Ci claims that Maoist China was a society of this kind. Material deprivation was not stigmatized but part of a communist ascetic ethos.¹⁸ Hence, concerns about the inability to afford goods and concerns about status differences due to wealth are distinct.

Contrast, then, two cases. In the first case, a homeless person is denied a good because of their actual inability to pay. In the second case, a homeless person is denied access to a shop despite having the ability to pay – perhaps because the shopkeeper is inaccurately deeming the homeless as unable to pay. The second case is direct socioeconomic discrimination and, in our view, prohibiting it would be justified. We do think that this is revisionary given that shops often use cues like homelessness to determine ability to pay, but we think that this is a welcome implication of our view. The first case is different. We can see this most clearly from a demeaning perspective. In a market society, failing to buy a good because one cannot afford it is simply the making of the market and does not express any lower status. By contrast, precisely because markets are supposed to be impartial being rejected, despite having the money to buy a good, signals that one is an outcast of society.¹⁹

The last observation also explains how the conduct of the shop owner in the second case contributes to social subordination insofar as it involves and contributes to stereotypes and stigma. But market pricing, by itself, does not subordinate the poor. Ci’s description of Maoist China makes this clear. Rather than market pricing, the distribution of money and the opportunity structures in society seems the most pressing cause of the subordination of the poor. Opportunity structures which fuel fear of downward social mobility or induce residential segregation are key contributors to subordination. Without these in place, status differences weaken.²⁰ This suggests that market pricing is not the culprit.

Freedom perspectives concur. The shopkeeper’s action in the second case denies the homeless person their deliberative freedom – homeless people need to factor in how they are perceived in society. But when it comes to market transactions, the unfreedom of homeless persons is caused differently, not by how the shopkeeper perceives the homeless person, but rather by the fact of a lack of money.²¹

Our second response is that the case of market pricing reminds us that not all classifications based on a certain trait amount to wrongful discrimination. In cases in which there is a strong connection between the job requirements and a protected trait, direct discrimination can be justified as a bona fide occupational requirement. Fair market prices could be exempted in a similar manner given that there is a strong justification for the market mechanism, and insofar as we just argued that charging a market price will not, in and of itself, create concerns about subordination or freedom.²²

The second reason also points to why market pricing will typically not amount to indirect discrimination (more on this shortly). The standards for justifying direct discrimination are at least as high as those for indirect discrimination. So if market pricing can be justified direct discrimination, then a fortiori it is also justified indirect discrimination.

We turn now to the impacts on the welfare state. This objection highlights a different virtue of class-based distinctions. Redistributive policies typically incorporate distinctions based on socioeconomic status. Progressive taxation involves a form of discrimination against the rich. The more you earn, the more you must pay in taxes – not just in absolute terms but also in relative terms. However, many theories of justice demand redistribution.²³ Would progressive taxation have to be declared an unlawful form of discrimination if discrimination laws forbade direct socioeconomic discrimination?

A first response points to the mismatch between income classifications and socioeconomic class as we understand it. We argued that class is broader than mere income levels. The tax system does not classify people by their social class but rather by their income level alone. Hence, it is not direct socioeconomic discrimination.

We wish to add a second reply for the following reasons. First, the critic may insist that because of the close connection between high income and high socioeconomic status, discrimination law should capture this case. Perhaps high income is a constitutive part of high socioeconomic status and should therefore be treated as part of it. (This is similar to arguments that hair discrimination is part of race discrimination – see also Section V.) Or perhaps we should treat this as indirect discrimination (more on this shortly). Secondly, on the beneficiary side of the welfare state, at least some programs target socioeconomic status in a thicker sense than mere income levels.

Our second response insists that its redistributive nature justifies progressive taxation. Affirmative action measures for race or caste, too, redistribute opportunities in access to higher education and public employment. We believe affirmative action is acceptable even though it distinguishes based on traits protected by discrimination law.²⁴ One option is to draft legislation such that it expressly permits redistribution similar to how the Indian and South African constitutions expressly permit affirmative action. Alternatively, one could define socioeconomic class asymmetrically as socioeconomic disadvantage. This would follow the model of disability and pregnancy discrimination. Thus, concerns about the market mechanism and the welfare state can be answered. The analogy remains intact.

III. THE REVISIONARY AND THE RADICAL CONCLUSION

If we combine the analogy with the widely accepted standard discrimination view, this leads us to the revisionary conclusion that anti-discrimination laws should be broadened to include direct socioeconomic discrimination (i.e., (C1)). The conclusion is revisionary in the sense that most jurisdictions currently do not prohibit direct socioeconomic discrimination.²⁵ The revisionary conclusion means that many current rules need to be changed. Employers could no longer reject applicants because of the socioeconomic class they were born into. Universities would have to adopt measures to avoid class bias on campuses. Housing societies would no longer be allowed to reject tenants willing and able to rent because they are presently poor. Shops could not deny service to homeless people who can afford to buy goods from them. These are welcome implications of our argument. Such changes in the law would help to reduce class bias and mitigate inequality of opportunity due to socioeconomic class.

So far our discussion has focused on direct socioeconomic discrimination, but we have said little about what direct – as opposed to indirect – discrimination is. Consider discrimination based on place of residence. Suppose an employer is wary of hiring any applicant who lives in working class areas. Is the employer discriminating against poor and working-class people? The employer might just as much discriminate against the few middle-class persons living in the area. The employer might say that they do not use socioeconomic class as their classification or rule but rather place of residence. However, the obvious concern is that applying a rule based on place of residence has a very similar effect as applying a rule based on socioeconomic class. We can capture this by saying that the rule based on place of residence is an example of indirect socioeconomic discrimination.

Specifying the precise distinction between direct and indirect discrimination can be tricky. Is a ban on headwear a direct exclusion of Muslim women and Sikh men or is it a neutral rule which disproportionately impacts them? For our purposes we adopt the following broad and ecumenical definition of indirect discrimination. Indirect discrimination is (i) a facially neutral practice, rule, policy, or act (hereinafter simply rule) which (ii) nevertheless imposes disproportionate disadvantages that (iii) cannot be justified as a proportional means to a legitimate end.²⁶ Direct discrimination involves a facially non-neutral rule and thus direct and indirect discrimination are mutually exclusive. Exactly where the line between direct and indirect discrimination is drawn will depend on how one interprets “facially neutral”. One interpretation is that the rule is facially neutral regarding the relevant protected trait if it does not classify people by that trait. Does the rule use race, caste, sex, and so on as a means to distinguish between people? One concern with this approach is that it would deem a “no turban at work” rule to be facially neutral even if adopted out of anti-Sikh bias. A second interpretation focuses on discriminatory intent. A rule is facially neutral if it is adopted without any discriminatory intent. While this would plausibly classify the “no turban at work” rule as non-neutral if it is based out of anti-Sikh bias, it would classify it as neutral if it is adopted because of a profit-maximizing motive. Perhaps customers are biased against Sikhs or turbans marginally impact productivity. A third interpretation, inspired by British law, is that a rule fails to be facially neutral once it excludes every member of the disfavored group. For example, a rule requiring hotel guests to be married would be non-neutral at a time when homosexuals cannot marry and thereby the rule excludes them entirely.²⁷ This rule would deem both turban rules as non-neutral but it also makes it hard to see why we should care much about the difference between complete and near complete exclusion.

Disproportionate disadvantages are those where the protected group is worse-affected than a comparator group. There must be “disparate impacts” for the group to be discriminated against. Lastly, not just any imposition of disadvantages on a group is discrimination. If it were, virtually any practice or rule could be indirectly discriminatory, since virtually any practice or rule will affect some group differently. The disparate impacts must be disproportional such that they are not outweighed by the achievement of a valuable end. Our third condition, inspired by the wording of the UK’s Equality Act 2010, covers what other jurisdictions spell out in specific legal doctrines like business necessity, relation to job performance, and the like. In line with our strategy to draw parallels to paradigmatic discrimination, we assume that the standards of proportionality for socioeconomic discrimination would mirror those used in paradigmatic discrimination.

Consider, then, a law prohibiting socioeconomic discrimination that prohibits not just direct discrimination but indirect discrimination as well. One reason for adopting such broad-scope prohibition would be that one accepts the unified view, and the revisionary conclusion. We just defended the revisionary conclusion, and we turn to arguments for the unified view in Sections IV and V. Before we do that, however, we want to illustrate ways in which the radical conclusion, is indeed radical.

Inheritance laws permit children in wealthy families to benefit by inheriting their parents’ property. A different, redistributive inheritance tax regime would make people of lower socioeconomic status noticeably better off. The substantial disadvantages the poor suffer under the present inheritance tax regime would make it difficult to justify the indirect discrimination it involves as proportional. Hence, the inheritance tax regimes currently in operation are disproportionately disadvantageous to them relative to an alternative radically redistributive scheme of legal regulation of inheritance and, thus, indirectly discriminatory.²⁸ Compare this with an inheritance regime which dramatically disadvantages women by setting conditions for larger inheritances which only few (if any) women can meet.

Next, schools in India provide free midday meals for children to tackle malnutrition. Eggs are an especially nutritious food. Including them in the meal plan would better serve the interests of the poorest children who are most at risk of malnutrition.²⁹ A decision to exclude eggs therefore disproportionately disadvantages poor children relative to a menu including eggs even though the policy is neither intentionally designed to do so, nor invokes any explicit distinction between poorer and better off pupils. It also would not be justifiable on the proportionality standards usually applied to indirect paradigmatic discrimination. This can be seen most easily when we change the example so that it involves disparate impacts by sex. If the policy was such that nutritional needs of women were served less well than the nutritional needs of men, it would be indirect sex discrimination.

Next, consider an example from employment law. Many employers regard unpaid internships as relevant job experience. However, applicants from poorer backgrounds will struggle to acquire job experience via unpaid internships since they cannot afford to work without being paid. If indirect socioeconomic discrimination was prohibited by employment law, employers would need to specially justify taking unpaid internships into consideration. The situation could become akin to the steps employers need to take to avoid pregnancy discrimination which indirectly discriminates against women. The mere fact that unpaid internships enable companies to gain higher profits would not count as a justification just as higher profits from reducing time of family leave would not count as a justification for pregnancy discrimination.

More examples could be marshalled ranging from challenges to private schooling, using educational prestige in hiring decisions, housing segregation by social class, or the Obamacare expansion mentioned in the introduction. Given that much of current practice is at stake, we need to look more closely at the unified view (i.e. (P3)). In the next two sections, we explore ways in which it can be supported.

IV. THEORIES OF DISCRIMINATION AND THE UNIFIED VIEW

It is not only the case that most jurisdictions prohibit both direct and indirect discrimination. The unified view of direct and indirect discrimination is also supported by most prominent theories of discrimination. In this section, we show that three of the four reasons we presented in support of the analogy (see Section I), also support the unified view.³⁰

Consider first an anti-social subordination view focused on group-level disadvantages. It holds that discrimination law should step in when a group is disadvantaged not only in material terms but also socially and politically. Insofar as it is the purpose of discrimination law to mitigate subordination of this kind, discrimination law should not be confined to direct discrimination. After all, this theory of discrimination shifts the focus: we are now thinking not just about individual interactions, but about broader social patterns. This is why, Owen Fiss who has proposed a theory of equality law along these lines, expressly attacks the singular focus on direct discrimination.³¹ We have already made the case, in our discussion of the analogy in Section I, that the poor are disadvantaged in a way that is captured by this concern about social subordination.

Relatedly, the argument also holds if we understand subordination in terms of deficits of social standing. In discussing gender and disability, Moreau introduces the term “structural accommodations” to refer to rules or policies which implicitly favor a dominant group. One example is physical infrastructure which implicitly favors able-bodied persons at the expense of disabled persons.³² Structural accommodations render disabled people “unnatural” or invisible. Disabled people are presented as deviations from the norm, less capable, and needing special treatment. Similar accommodations exist in connection with socioeconomic class. The physical infrastructure in most of the United States makes owning a car a necessity. That is structurally unaccommodating of poor people, who cannot afford a car and then struggle to get to work, and so on. When poor Americans are disproportionately late for work, then this serves to reinforce stereotypes of poor people lacking a proper work ethic.

Consider next freedom-based understandings of discrimination. Earlier we discussed an influential theory of this sort, due to Khaitan, which revolves around the thought that the point of discrimination law is to provide everyone with the basic freedom to lead a good life.³³ This approach to discrimination law privileges indirect discrimination over direct discrimination – and thus has the implication that the unified view, is true, because the reasons favoring the prohibition of direct discrimination not only apply to the prohibition of indirect discrimination but do so with greater force than they do to direct discrimination. The key concern of discrimination law is the disparate impact of patterns of action on individual freedom. In this sense, the best justification for discrimination law, for Khaitan, embraces the unified view.

Lastly, recall the idea of deliberative freedom – the freedom not to have to treat one’s race, caste, or sex as a cost in deciding how to lead one’s life.³⁴ While many infringements of this freedom involve direct discrimination, some are caused by indirect discrimination. Consider admissions tests for higher education or employment which in part test the cultural cues of dominant groups. These cues are a subtle reminder to members of oppressed groups of where they supposedly do (not) belong. For those members, they feature as costs in the decisions they make about what career to pursue, and whether to prepare for the tests. For students of working-class background, it is not only the admission stage of higher education that imposes costs that relate to socioeconomic origin. These costs can be subtle effects of the way universities are run to fit specific social backgrounds even in the absence of anyone’s mentally representing them as working-class students, and even in the absence of any rules that explicitly pick out working-class people. They can include facts about which cultural forms of expression are valued on campus, how options for out-of-class support are structured based on the student taking the initiative, and so on.³⁵

We have considered three prominent reasons for prohibiting indirect discrimination by law. The first drew upon ideas of subordination and inequality, the last two on ideas of freedom. All offer powerful support for the unified claim from different theoretical angles. Thus, anyone who rejects the unified view must also reject a wide and diverse range of views on why discrimination should be prohibited.

V. DISCRIMINATION BY PROXY TRAIT

Some will remain reluctant to accept the implications of outlawing indirect discrimination on grounds of socioeconomic class. Moreover, not all reasons offered for prohibiting direct discrimination by law apply with equal force to indirect discrimination. Someone resisting the radical conclusion could adopt a theory of discrimination law which only applies to direct discrimination.³⁶

Even if we grant that only direct discrimination is a genuine form of discrimination, there is a reason why indirect discrimination law developed. A key point for indirect discrimination law was Griggs v. Duke Power. Duke Power had overtly racist employment practices throughout the 1950s. On the same day as the Civil Rights Act came into force in 1964, Duke Power changed its policies and introduced new employment tests for those applying for posts in their better paid departments that, despite making no reference to race, Black applicants were significantly less likely to pass. Duke Power was no longer directly discriminating against Blacks but had achieved almost the same result.

The employment tests used by Duke Power were proxies for race. Today, more and more concerns about discrimination relate to practices that invoke proxy traits. In particular, many proxies for socioeconomic class are concerning. One reason for this is the fuzzy and messy definition of class. In many cases of socioeconomic discrimination what we really have is differential treatment based on the more specific traits that jointly make up socioeconomic status or origin. There is accent discrimination, place of residence discrimination, discrimination against low-brow culture, and so on. This does not entail the reductionist view that there really is no such thing as socioeconomic discrimination. Race discrimination works in similar ways. Social constructivist understandings of race also point to a fuzzy number of traits that make up race (e.g. hair texture, skin color) which sometimes overlap with class (e.g. accent).³⁷ What we should say in both cases is that discrimination based on those more fine-grained traits is discrimination based on class (or race).

However, there is still a lingering concern that many of these traits only imperfectly map onto class. There are, after all, people from a working-class background who enjoy high-brow culture. Focusing only on direct discrimination requires us to make a sharp distinction between traits so closely associated with class such that they are constitutive and traits which are not. Yet plausibly traits associated with class form a continuum rather than two discrete categories. This leads to two problems. First, it seems arbitrary to draw such stark distinctions between such traits. Second, doing so allows agents to discriminate on proxy traits rather than constitutive traits and achieve almost the same result. A blanket rejection of the idea of indirect discrimination would make socioeconomic discrimination law toothless.³⁸

Let us take stock of the argument of this section. We have examined the suggestion that we have should give up on indirect discrimination law altogether (to avoid the radical implications of the radical conclusion). In a nutshell, we have argued that this is not an appealing route to take. Admittedly, this does not amount to a positive argument for the unified view, but it supports it indirectly by showing how the alternative is flawed.

VI. INSTITUTIONAL LEGITIMACY AND COMPETENCE

If one accepts our argument up until this point, then one might be tempted to accept the radical conclusion, i.e. (C2). We have supported all premises of the core argument. So why not accept its conclusion? In this and the next section we explore objections to the radical conclusion which explain why its radicality is troublesome – if not in principle, then at least in practice. If these objections succeed, then one of the premises in the argument in its favor must be false even if our examination failed to exhibit which one(s).

One such response notes that the examples of indirect socioeconomic discrimination we listed in Section III all favored policies championed by the political left. With indirect socioeconomic discrimination law in place and proportionality standards akin to those used for indirect paradigmatic discrimination, courts could declare as unlawful the exclusion of eggs from midday meals, the insufficiently redistributive system of inheritance taxes, and other social policies. In jurisdictions with judicial review of legislation, this would allow courts to overrule legislation which indirectly discriminates against the poor. In all jurisdictions, it would allow courts to overrule administrative actions – like the exclusion of eggs in midday meals. Either way, adopting indirect socioeconomic discrimination law would shift the balance of power away from legislatures and governments towards courts and would to some extent remove such policies from the scope of democratic contestation. While such a law would (presumably) be passed by an elected parliament, our point is that the interpretation of this law would be far-reaching and such interpretation would have to be done by courts. Yet many think that economic and social policies are not legitimately crafted (or corrected) by courts. In the 1960s, when it briefly appeared that the US Supreme Court would include socioeconomic discrimination under the ambit of the Equal Protection Clause, Justice Harlan articulated this concern noting:

It was not too long ago that Mr. Justice Holmes felt impelled to remind the Court that the Due Process Clause of the Fourteenth Amendment does not enact the laissez-faire theory of society. … it is appropriate to observe that neither does the Equal Protection Clause of that Amendment rigidly impose upon America an ideology of unrestrained egalitarianism.³⁹

Accepting the radical conclusion would mean that discrimination law becomes deeply entangled with judgments of distributive justice. However, the maintenance of a degree of distance between anti-discrimination laws and distributive justice allows people of very different political ideologies to come together in their condemnation of discrimination. The expansion of concerns about egalitarian justice into the sphere of anti-discrimination law makes those concerns acceptable only to a narrower group of political philosophies. Call this the legitimacy objection.

We have two responses. First, a prohibition on indirect socioeconomic discrimination would not legislate a particular view of distributive justice. Indirect socioeconomic discrimination is tied to group membership, whereas most approaches to distributive justice focus on the entitlements of individuals. Indirect socioeconomic discrimination law is about class politics, not individual entitlements.

Second, the legitimacy objection does not say what distinguishes socioeconomic class from race, caste, or sex. Current indirect sex discrimination law requires courts to implement (gender) equality policies and be, in a way, explicitly feminist. Why should we be more lenient in the case of socioeconomic discrimination than we are in the case of race or caste discrimination – especially if, morally, socioeconomic discrimination is no different from paradigmatic discrimination as we argued in Sections I and IV? If socioeconomic discrimination is as omnipresent in society as the legitimacy objection implies, do we not have even more reason to be concerned? Would we change our minds about the importance of using the law to eliminate race or caste discrimination if we found out that tackling them would require much more extensive intervention in society than we had anticipated because so many transactions today are in some way racist and casteist?

This is not just a thought experiment. A slightly weaker version of the radical conclusion can be derived even with paradigmatic discrimination alone. In many societies socioeconomic status correlates strongly with one of the traditional traits inherent in paradigmatic discrimination. Race and caste are the obvious examples. Most of the examples we have introduced in this article are practices or policies, with disproportionate impacts on, not only on poor people, but also Blacks or oppressed castes, simply because these groups are more likely to be poor or to have been born poor.⁴⁰ Thus, many of these policies amount to indirect race or caste discrimination as well.

Hence, a slightly weaker version of the radical conclusion can be defended without appealing to socioeconomic discrimination. We are not convinced that this point should make us rethink and soften our stance on tackling indirect race or caste discrimination. But then why should we be impressed by a similar appeal to soften our stance on tackling indirect socioeconomic discrimination? So much for the legitimacy objection.

A second concern about the radical conclusion, closely related to the legitimacy objection, is that even if we grant that courts can in principle legitimately judge on matters of educational policy, tax policy, and the like, judges are not experts in social policy. The adjudication of a case of indirect socioeconomic discrimination would require not only knowledge of the distributive effects of a social policy but also a grasp of the possible alternatives. Democratic politics, for all its flaws, allows a broader range of opinions to be heard. It is a strength of democracy that it brings out the epistemic virtues of inclusive deliberation and decision-making. This leaves judges with a dilemma. If, on the one hand, judges interpret the demands of indirect socioeconomic discrimination law stringently, then they stand accused of a competence deficit relative to democratic politics. But if, on the other hand, judges, for this reason, give much leeway to actors to justify indirect socioeconomic discrimination, then the law risks becoming toothless. Arguably, concerns about competence explain why socioeconomic discrimination law even where it exists, has hardly made any impact. Call this the competence objection.

We have three replies. First, one can interpret this concern not as an argument against laws prohibiting socioeconomic discrimination, but as an argument for the training of legal professionals to acquire the relevant expertise regarding the effects of social policy.⁴¹ Second, the dilemma might be somewhat overstated. Even if a court is somewhat deferential to democratic politics when it comes to social policy, there are still instances of socioeconomic discrimination which are easier to capture. Third, it is again unclear what distinguishes socioeconomic class from race, caste, or sex here. Why should the courts be any better equipped to tackle structural racism? Structural racism is an extraordinarily complicated phenomenon, and a legal perspective is just one perspective among many. The competence objection applies mutatis mutandis to paradigmatic traits, too, and yet we suspect that few people who would press the competence objection against socioeconomic discrimination laws would offer it to reject laws forbidding race, caste, or sex discrimination.

To conclude, we see the force of the legitimacy and competence objections. There are some considerations which mitigate them and, importantly, it appears that both objections apply to some extent to indirect paradigmatic discrimination law, too.

VII. COSTS OF INDIRECT SOCIOECONOMIC DISCRIMINATION LAW

The legitimacy and competence objections concern discrimination law as it applies to state action. We now turn to a worry about the costs of outlawing socioeconomic discrimination for private actors. At present, employers must absorb the costs of avoiding discrimination in certain scenarios. For example, they must ignore the cost of parental leave in hiring decisions. They also need to absorb the cost of making workplaces accessible to disabled people. That said, by and large, employers are free to choose employment practices that maximize profits. Indirect socioeconomic discrimination law would disrupt this balance. As Joseph Fishkin puts it:

The problem is that almost everything has a disparate impact based on class: there are very few employment practices of any kind that make distinctions among employees or applicants that do not have some significant class-based disparate impact. Therefore, a body of disparate impact law focused on class would be tantamount to a legal rule that nearly all employment practices must meet the business necessity/job relatedness test.⁴²

If the radical conclusion is accepted, beside meeting existing demands, employers will now be required by law to justify hiring practices that give undue weight to, for example, educational prestige and background. Employment discrimination law will become much more expensive for employers as its scope expands.⁴³

A couple of considerations mitigate this problem. First, the costs of discrimination law to businesses are often smaller than they appear to be, because one’s competitors are equally obligated to abide by the same laws: there is no competitive disadvantage here.⁴⁴ Second, more inclusive practices come with long-term benefits for employers. They help reduce class bias, for example, and thus help employers recruit from a larger pool of talent.

But, arguably, these considerations do not put the concern about costs to rest. We see three possible responses to the problem of costs, none of which is fully satisfactory. The first is to bite the bullet and insist that the employers’ profits can reasonably be sacrificed for the greater good of a less class-ridden society. However, the additional costs are significant. Because of the way skills are tied up with one’s family background, adding socioeconomic origin to employment discrimination laws is a radical step. It creates costs substantially greater than those involved in, say, indirect sex discrimination law.

Second, we could take this problem to show that indirect socioeconomic discrimination laws should not be implemented. The difficulty with this is that we still have not identified what, other than actual costs, makes socioeconomic discrimination relevantly different from paradigmatic discrimination law. Suppose it were to turn out that caste discrimination law is much more costly for employers than had been thought initially. Would that persuade us that we ought to re-think the importance of tackling caste? Caste disadvantages are very deeply tied up with the acquisition of skills and the traditional link between caste and occupation continues to exert strong influence.⁴⁵ Perhaps, then, Fishkin’s observation would apply with equal force to indirect caste discrimination law as well. It seems highly counterintuitive to think that since caste has extraordinary influence over people’s lives, we should adopt a less ambitious policy on it.

The point can be put in another way. Imagine a society in which educational disadvantages based on socioeconomic class have been much reduced: the link between acquiring skills and one’s family background has been weakened considerably, but not eliminated. And despite the more egalitarian education system, class bias persists, and members of the working class are subjected to social stigma and public ridicule. In this society indirect socioeconomic discrimination law would be less costly because the relation between skills and socioeconomic origin is weaker. However, the key rationale for socioeconomic discrimination law would continue to apply. Hence, the argument against indirect socioeconomic discrimination law – at least when it comes to employment – would no longer be as powerful. Our question is this: Does it not appear perverse that it is precisely the fact that the poor are disadvantaged less that now persuades us to protect them better in law? Conversely, why should the fact that class disadvantages are worse be a reason to give up the idea of protecting the poor from those disadvantages through discrimination law?

The third reaction to the problem of excessively costly indirect socioeconomic discrimination law is to lower the proportionality standards for indirect discrimination. If socioeconomic discrimination is on a par with other forms of paradigmatic discrimination, and concerns about employers’ costs are weighty, then one answer is to lower the costs across the board. This answer has the advantage of treating socioeconomic discrimination no different than paradigmatic forms of discrimination. But if we add socioeconomic discrimination to existing discrimination laws, then the only way to prevent costs from rising is to give employers more leeway when it comes to justifying disparate impacts. This would lessen the protection discrimination laws offer to presently protected groups, something we are not content with either.

None of the three answers are fully satisfactory. This speaks to our claim that there is a puzzle here. We remind the reader that the main aim of this article is to reveal a deep incoherence in views on discrimination that most of us hold. The radical conclusion follows from plausible premises that have withstood examination. But the radical conclusion is not only strongly counterintuitive – even to people who are strongly repulsed by paradigmatic forms of discrimination – but faces presently unresolved objections.

VIII. COMPETITION WITH OTHER TRAITS

The third answer reviewed above brings us to the last reason we examine for rejecting the radical conclusion – not as false, but as politically undesirable. Sometimes a focus on socioeconomic disadvantage is seen as in competition with appropriate emphasis on the disadvantages of race, caste, and sex. Specifically, the immediate response of some left-leaning academics to the present article is to worry that it reflects, or may support, a right-wing populist view that detracts from the seriousness of race, caste, or sex discrimination.

We understand this reaction. However, our view is not that people worry too much about paradigmatic forms of discrimination, but rather that they worry too little about its socioeconomic counterpart. We want to make two points here. First, in the societies where it is applied, discrimination law has already moved us beyond, and away from, the most horrific forms of discrimination, and this has strengthened the case against the latter. Writing in the 1970s, Owen Fiss worried that the extension of discrimination law to women and religious minorities would be “potentially antagonistic” vis-à-vis the paradigmatically protected group of Black Americans.⁴⁶ Thankfully, this extension did not dilute protection for African Americans. The general lesson is that discrimination law tackles different faces of discrimination, with different levels of intensity and seriousness. Historically, efforts to tackle the less serious forms of discrimination do not appear to have resulted in trivialization of the more serious forms in the eyes of most people.

The second point is about the synergies between socioeconomic class, on the one hand, and race (e.g., in the United States, South Africa, and Brazil) and caste (e.g., in India, Nepal, and Pakistan), on the other. Often the two forms of discrimination are linked: poverty is disproportionately concentrated in oppressed racial and caste groups, and intersectional discrimination occurs as a result. If the law is to grasp such intersectional discrimination, it will be helpful to have socioeconomic status, or origin, categorized as a protected trait.

IX. CONCLUSION

We have examined two of the three premises of the core argument: the analogy (P2), and the unified view (P3). We took the standard view that direct race/caste/sex (paradigmatic traits) should be prohibited by law (P1), as uncontroversial. All three premises, in our view, capture something intuitive and plausible about discrimination. True, there are ways to question the analogy and ways, also, to dispute the unified view. But we have set these out and defended them. This leaves us with the revisionary (C1), and – more importantly – the radical conclusion (C2). Perhaps anti-discrimination law should indeed become radically egalitarian and intertwined with distributive questions. So, perhaps, after all, the cases Patient and Obamacare should both be regulated by the same law. Why should we not simply accept this? We presented some pressing concerns that are yet to be satisfactorily answered. As we have indicated, none of the routes of escape from the puzzle we have sketched is entirely satisfactory. There is a lot to be said for the standard discrimination view, the analogy, and the unified view, yet there is also a lot to be said against the radical conclusion. The puzzle forces us to reconsider the interaction between two aspects of egalitarian thought: distributive justice and discrimination.

COMPETING INTERESTS

The authors declare that they have no competing interests.