Indirect Discrimination Is Not Necessarily Unjust
This article argues that, as commonly understood, indirect discrimination is not necessarily unjust: 1) indirect discrimination involves the disadvantaging in relation to a particular benefit and such disadvantages are not unjust if the overall distribution of benefits and burdens is just; 2) indirect discrimination focuses on groups and group averages and ignores the distribution of harms and benefits within groups subjected to discrimination, but distributive justice is concerned with individuals; and 3) if indirect discrimination as such is unjust, strict egalitarianism has to be the correct account of distributive justice, but such egalitarianism appears vulnerable to the leveling down objection (whether decisively or not), and many theorists explicitly reject strict egalitarianism anyway. The last point threatens the position of liberals who oppose indirect discrimination but think significant inequalities can be just.
In most Western countries many forms of direct discrimination are illegal. Employers can be fined and required to pay compensation if they reject applicants on grounds of race, gender, religion, or sexuality. Not only are such actions illegal. Most people consider direct discrimination on these grounds unjust across a wide range of contexts. I write “a wide range of context” and not “all contexts”, because many do not believe it is unjust if, say, a film director “making a film about the lives of blacks living in New York’s Harlem” treats applicants differently on the basis of race (See Singer 1978, p.188).1
Initially, many hoped that once we got rid of direct discrimination, inequalities of race, and gender, and so on, would wither away, but clearly the legal prohibition of direct discrimination has not had this result. This is where indirect discrimination enters into the picture. The famous 1971 US Supreme Court ruling—Griggs vs. Duke Power—confirmed that a rule or practice can be illegal when it is “fair in form, but discriminatory in operation”—or, to put it differently, indirectly discriminatory (Fredman 2011, p. 178; Connolly 2011, p. 152; Griggs v. Duke Power 1971). In the case at hand an employer, Duke Power, “instituted requirements of high school education and satisfactory scores in an aptitude test as a condition of employment or transfer. The same test was applied to all candidates, but because black applicants had long received education in segregated schools, both requirements operated to disqualify black applicants at a substantially higher rate than whites” (Fredman 2011, p. 178). Since the relevant requirements were not needed to ensure satisfactory levels of performance, the company was ordered to abolish the requirement and to address the underrepresentation of Afro-Americans on its staff.
The 2009 Supreme Court ruling in Ricci v. DeStefano has to a large extent reversed the Griggs vs. Duke Power ruling. However, the idea that acts, practices and rules can be indirectly discriminatory, and therefore unjust, as a result of their differential effects, and in the absence of any intention to exclude members of any group, has had a huge impact; and many legal codes now recognize indirect discrimination as a prohibited category along with direct discrimination (Ricci v. DeStefano 2009; See also Selmi 2006). For instance, various European Court of Human Rights rulings have embraced the view that indirect discrimination falls under the European Convention on Human Rights. Also, EU Council directives mandate implementation of the principle of equal treatment irrespective of racial or ethnic origin in part through the prohibition of direct as well as indirect discrimination. (DH v. Czech Republic 2008; see also Shanagan v. UK 2014).2
While many liberals favour the legal prohibition of indirect discrimination, it raises a number of thorny issues. One is about the list of protected groups that most such prohibitions involve (see note 5). Why are the groups mentioned above on the list? Consider age. In some contexts rules that disadvantage certain age groups seem just. Rules of organ transplantation prioritizing the needs of young patients, who have enjoyed few worthwhile years of life, over those of older patients might be an example—e.g. rules of organ transplantation prioritizing the needs of young patients, who have enjoyed few worthwhile years of life, over those of older patients—seem just (Kappel and Sandøe 1992, pp. 297-316). Again, why are certain groups, such as the obese, or those on low-incomes, absent from the list?3 Certainly, people with obesity or on a low-income are seriously disadvantaged by various rules and practices that seem—even are—fair in form.
These questions are hard to answer. However, they arise in connection with both direct and indirect discrimination, and my focus here is on questions specifically about the latter (Lippert-Rasmussen 2013, chapter 1). I begin, in Section II, by expounding an Altmanesque definition of indirect discrimination with the aim of presenting three core challenges to the view that indirect discrimination is unjust as such. Section III focuses on the distinction between local disadvantage, e.g. underrepresentation of certain groups among CEOs, and global disadvantage, e.g. disadvantage in terms of the overall of resources. This distinction gives rise to the local-global disadvantage dilemma: Either accounts of indirect discrimination concern the former, in which case indirect discrimination is not unjust as such, or they concern the latter, in which case they are radically revisionist. Section IV notes that mainstream theories of indirect discrimination determine disadvantage on the basis of group averages. The gives rise to the challenge from group averages: In the light of intragroup inequalities, indirect discrimination is not always preferable, justice-wise, to its absence. Section V shifts the focus from disadvantage to disproportionality—both essential components in indirect discrimination—and distinguishes between two interpretations thereof: One that compares inequalities between groups under situations with and without indirect discrimination—the relativized view—and one that compares how well off the group being subjected to indirect discrimination is under situations with and without indirect discrimination against it—the absolute view. Section VI shows that only the former view fits standard conceptions of indirect discrimination. However, this implies—and that is my third and final challenge—that the view that indirect discrimination as such is unjust is vulnerable to the so-called levelling down objection and, thus, that to endorse this view one has to reject this objection. I conjecture that many who find indirect discrimination unjust will find this an unwelcome implication of their view. After all, it is commonly assumed that one can consistently oppose indirect discrimination without subscribing to strict egalitarianism. Section VII responds to three objections to my levelling down challenge and makes some cautious remarks about its limitations. Section VIII concludes by exploring the practical implications of the views defended here, i.e., that because indirect discrimination is not unjust as such acts that indirectly generate group disadvantages need not be unjust and, thus, might be such that they should be legally permitted.
Political philosophers have paid surprisingly little attention to the question why discrimination is unjust compared to other political charged questions such as “What makes wars just?” and “Should abortion be legal?” In fact, I do not think that there is a reasonably well established, or well-expounded view of what makes discrimination unjust (when it is). Accordingly, this article should not be seen as a refutation of such a view, but more as an important new step into under-theorized territory in political philosophy. That being said, the view that discrimination as such, and by implication indirect discrimination which after all is a species of discrimination, is unjust is common. For instance, James W. Nickel writes: “Discrimination is morally wrong because its premise that one group is more worthy than another is insulting to its victims, because it harms its victims by reducing their self-esteem and opportunities, and because it is unfair” (Nickel 2000, p.214). Similarly, Lena Halldenius uses the term “discrimination” such that “[w]hen an action has been correctly described as an instance of discrimination, it has at the same time been correctly described as unfair” (Halldenius 2005, p. 456). 4 In my view, the assumption that discrimination as such is unjust deserves closer scrutiny. This is true of direct as well as indirect discrimination, but, as already noted, here I restrict my attention to indirect discrimination and it is more plausible to deny that indirect discrimination, as opposed to direct discrimination, is unjust as such, because the latter involves treatment that is unfair or involves objectionable mental states irrespective of its consequences (See however Lippert-Rasmussen 2013, pp. 103-189).
While it is not really necessary to mount my three core challenges, throughout this article I shall assume that if a certain act is unjust that constitutes a reason for the moral desirability of the act being legally prohibited. This assumption is quite weak and is acceptable to a wide range of theorists. First, it does not rule out there being non-justice based, potentially overriding, reasons for or against legal prohibitions of acts, e.g. that they promote or reduce general welfare. Second, on many views injustice is cashed out in terms of violation of rights—in the case of discrimination: human rights—and it is commonly assumed that the law ought to prohibit (human) rights violations. Finally, legal moralists believe that the fact that an action is morally wrong is a reason to prohibit it. While some endorse legal moralism, many reject it, but even most of those, who do, accept that the subset of morally wrongful acts that involve injustice ought, morally speaking, to be legally prohibited.
II. Indirect discrimination defined
To determine whether indirect discrimination as such (henceforth I take this qualification for granted) is unjust we need to know what it amounts to, since, presumably, if it is unjust as such (henceforth I take this qualification for granted), it is unjust in virtue of features that necessarily belong to it.5 In particular, we need to have a clear view of the respects in which indirect discrimination differs from direct discrimination. In an encyclopedia entry on discrimination, Andrew Altman rightly notes that there is no agreed test, or criterion, of indirect discrimination (Altman 2011). Still, drawing on Altman’s work I propose the following definition, one that fits a number of existing characterizations of indirect discrimination quite well (Cf. (Halldenius 2005, p. 459):
A policy, practice or act is indirectly discriminatory against a certain group if, and only if: 1) it neither explicitly targets nor is intended to disadvantage members of the group (the no-intention condition); 2) it disadvantages members of the group (the disadvantage condition); and 3) the relevant disadvantages are disproportionate (the disproportionality condition).6
All three conditions point to differences between direct and indirect discrimination. The no-intention condition captures the core difference—the idea being that a company, say, could indirectly discriminate against women even if it is neither explicitly targeting them (e.g. in job advertisements that invite applications from men only) nor intending to disadvantage them. (There is a different and non-intention related sense of “indirect”, which should be distinguished from the sense of “indirect” I expound here: this is the sense in which using a certain proxy (e.g. being taller than 1.85 meters) for pursuing one’s aim (excluding women) is indirect. Discrimination that is indirect in this sense is direct discrimination in my sense.)
The disadvantage condition also captures a difference between direct and indirect discrimination. To directly discriminate one has to treat the discriminatee of one’s actions disadvantageously in some way. However, in some circumstances one can do this without the outcome of one’s actions actually being disadvantageous to the discriminatee. Suppose a homophobic employer initially decides to hire a straight applicant rather than a better qualified gay applicant, but is then forced to offer the job to the latter because the former withdraws his application. The gay applicant was subjected to direct discrimination—the employer initially decided not to hire him on account of his sexuality—even if, as it so happened, the relevant outcome was not harmful for him. More generally, while indirect discrimination is tied to the outcome of the allegedly discriminatory process, direct discrimination requires only that a person be subjected to disadvantageous treatment. (Here I set aside here outcome-focused conceptions of direct discrimination according to which cases such as the one I described above involve attempted, but unsuccessful, direct discrimination.) (Lippert- Rasmussen 2013, p.18; Gardner 1996; Connolly 2011, p. 155)
The disproportionality condition reveals a third difference between direct and indirect discrimination, for neither it nor any similar condition must be satisfied in cases of direct discrimination. Suppose there is some morally good reason to engage in direct discrimination. For example, we are in a country with a conservative, sexist majority that will predictably descend into civil war unless the established church directly discriminates against women when appointing people to religious offices. The interest in avoiding civil war morally outweighs the interest in sexual equality in the process of making church appointments. Here women are directly discriminated against when they are not hired for the relevant positions. Yet, because the case does not satisfy the disproportionality condition, a similar, but indirect case would not involve discrimination.
While this account of indirect discrimination can be improved upon in various ways, it suffices for our purposes, (Lippert-Rasmussen 2013, chapter 2) and I want now to tackle some issues raised by the question whether indirect discrimination is unjust. In doing so, I shall disregard the no-intention condition and focus on conditions 2) and 3). It is possible that indirect discrimination is unjust because it satisfies the disadvantage or the disproportionality condition. However, it cannot be unjust, because it neither explicitly targets, nor is intended to disadvantage, members of a certain group. After all, if targeting or intending to disadvantage makes a moral difference, justice-wise, it makes a difference to the worse, not the better.
III. Local v. Global Disadvantage
I begin with the disadvantage condition—the notion that indirectly discriminatory practices always disadvantage the group discriminated against. This condition is in need of clarification in two dimensions at least, and in ways that challenge the view that indirect discrimination is unjust. First (I will come to the second clarification in Section IV), a practice may disadvantage members of a certain group locally, or globally, as it were. If, on the one hand, disadvantage is understood locally, our concept of indirect discrimination is non-revisionist, but indirect discrimination is not unjust. If, on the other hand, disadvantage is construed globally, indirect discrimination is possibly unjust, but the emerging notion of indirect discrimination is also highly revisionist. This is the local-global disadvantage dilemma.
To see what the distinction between local and global disadvantage amounts to, imagine that language tests used by humanities faculties to select students tend to result in the admission of fewer immigrants. However, instead of being admitted to the humanities they seek admission at law schools, medical schools, engineering schools, and the like, where, as a result they are overrepresented. Suppose also that as a result they end up living lives which are better than the lives of non-immigrants. Here, a formally neutral rule disadvantages immigrants locally: they find it harder to meet the language test and struggle to gain admission to the humanities faculty. But the same rule advantages the immigrants globally: they end up being better off overall than other members of society. Whatever objectionable features the relevant admission rule has in virtue of its impact on global distribution, injustice to immigrants cannot be counted among them.
When a rule or practice is criticized as indirectly discriminatory, the focus is on local, not global, disadvantage—e.g. the disadvantage reflected in the fact that women are underrepresented among professors or CEOs. This may reflect our tendency, when raising complaints about indirect discrimination, to become exercised by local disadvantages that we take to contribute to a connected global disadvantage. This is why, presumably, although there are some rules and practices that place men at a local disadvantage (think of parental access to children following divorce), it is rare to hear of indirect discrimination against men (See, however, Sullivan 2004).
In the moral assessment of indirect discrimination the distinction between local and global disadvantage becomes important. Many would say that justice is concerned with the distribution of global benefits and burdens. On this view, the fact that some people are better off than others in some particular dimension—say, they have a higher income—can be counterbalanced by the fact that they are worse off than others in another dimension—they have longer working hours and less autonomy in their jobs. Undoubtedly, there is something right in the view that justice is concerned with the distribution of global benefits and burdens; it would be odd to hold that it makes no difference, from the point of view of justice, whether local disadvantages counterbalance or accentuate one another. Against this view, it might be argued that it would be odd for an indirectly discriminating employer to get off the law’s hook simply because members of the group which she disadvantages, say, in terms of employment are advantaged in terms of other local goods, but in ways that are beyond this employer’s control. However, insofar as disadvantaged groups are identified not relative to each individual employer but, say, relative to the job market as such, it is any case true that individual employers are held responsible in part on the basis of facts that they do not control.
Some people, notably Michael Walzer, have defended the view that there are different spheres of justice, and that justice requires the goods within each sphere to be distributed according to criteria reflecting the nature of the relevant goods. For example, medical services should be distributed according to need, and places at universities according to merit (Walzer 1983).7 On a hybrid, Walzerian view where justice requires each good to be distributed according to its cultural meaning and equality of global advantage, indirect discrimination could be unjust because it results in local disadvantage. Obviously, alternative hybrid views of the way local disadvantage matters can be envisaged, but Walzer’s view is certainly the best known.
In response to the Walzerian position here, I note, first, that complaints about indirect discrimination often relate to disadvantages which, even on Walzer’s view, involve local disadvantage within a certain sphere. If, for instance, certain rules and practices lead to worse health outcomes vis-à-vis a particular disease for women from a Walzerian perspective, this would qualify as a local inequality in the treatment of a particular medical need, and yet it is compatible with the sphere of health as a whole being just in the sense that, globally speaking, health care is distributed according to need overall. Second, on Walzer’s view the social meaning of many goods implies they should not be distributed equally—e.g. admission to university should be based on merit. Accordingly, on Walzer’s view one group might be worse off than others in terms of the distribution of a particular good without this distribution violating the social meaning of the good, in which case it could not involve injustice, let alone unjust, indirect discrimination. Hence, one cannot build an account of the injustice of indirect discrimination on Walzer’s theory of justice. This completes my presentation of the local-global disadvantage dilemma.
IV. Group Averages and Intragroup Inequality
Let us now turn to the second dimension in which the notion of group disadvantage needs to be clarified. The basic issue here is that members of a group may be affected differentially by rules that, on average, (dis)advantage members of the group. Consider a test used to appoint senior managers which places a premium on being assertive, and assume it has following features. On average, women tend to score less well than men on it. Accordingly, despite equal numbers of men and women applying, more men than women are hired. Women and men vary in terms of how assertive they are. Some women are more assertive than most men, and some men are less assertive than most women. So, while it might be true that the test in question disadvantages the women, the sub-group of especially assertive women may actually benefit from the rule (some would not have been hired had the test not been used) and the sub-group of especially unassertive men are harmed by it (some would have been hired had the test not been used). Given these features, the charge that the test indirectly discriminates against women and in favour of men seems insufficiently specific. Why not say that it indirectly discriminates against the sub-groups of unassertive people, men and women?8 In itself this is an interesting question, but even if it can be answered in a principled way, there is another more worrying problem.
A rule, which, on average, disadvantages members of a certain group relative to another group, may in fact benefit most members of the group modestly provided that a few members are harmed a great deal (Cf. Doyle 2007). It may also be true that the few members who are seriously harmed by the rule are much better off than the rest in the absence of the rule. By way of illustration:
|5% best off men||All other men||Men average||5% best off women||All other women||Women average|
|Benefits under Rule I||490||90||110||130||110||111|
|Benefits under Rule II||100||100||100||120||120||120|
On average, Rule II makes men worse off, but it also reduces the inequality between most men and most women, and it reduces male intragroup inequality since the harm it causes relative to Rule I falls on the 5% best off men.
Again, in response to these facts it is seriously inadequate simply to say that Rule II indirectly discriminates against men—for two reasons. First, in the absence of Rule II most men would be even worse off relative to members of other groups, so, given a plausible measure of the injustice of overall inequality, Rule II may in fact reduce unjust intergroup inequality (Temkin 1993, pp. 19 – 52). Hence, if we feel indirect discrimination is unacceptable because we find group inequality objectionable, this is a case of indirect discrimination we should not object to. Second, Rule II reduces intragroup inequality between men, and if we think that justice is more concerned with the plight of badly off men than with that of privileged men, it is not clear that we should prefer Rule I over Rule II from the point of view of justice, small benefits to many worse off people may outweigh substantial harms to a few better off people even if the total sum of benefits is greater in the outcome that favours the better off. Hence, if “indirect discrimination” picks out an injustice (or, at any rate, a prima facie injustice), then, despite the fact that Rule II makes men worse off on average it should not qualify as a case of indirect discrimination. This is the challenge from group averages.
Admittedly, this challenge assumes, first, that views of justice that focus on inequalities between groups ignore intragroup inequalities between individuals by favouring some trade-offs of greater inequality between individuals for less inequality between groups and, second, that this renders such views implausible (Holtug and Lippert- Rasmussen 2007, pp. 6 – 7). I find both claims plausible. Indeed, in my example Rule II seems to involve less objectionable inequality than Rule I despite that, on a view that focuses on group averages, it is the former which involves more indirect discrimination.
Admittedly, if disadvantages tend to cluster, the gap between local and global disadvantage explored in Section III will rarely arise (Wolff and De-Shalit 2007). Similarly, if it almost never turns out that on average a rule disadvantages, say, an oppressed minority even though most of its members actually are better off living under the rule than they would be in its absence, the challenge from group averages will almost never be a practical problem. (Of course, in the Griggs v. Duke Power case African-Americans with a high school degree were in one respect better off with Duke Power’s rules of promotion than they were without it, since they faced no competition from fellow African-Americans without a high school degree.) On these assumptions, it is often best from the perspective of a political reformer to disregard such cases.9 However, if we look at indirect discrimination from the perspective of the fundamental principles of justice—principles which are required to apply to all scenarios, and not merely to those that are actual or likely—we cannot ignore the local-global advantage dilemma and the challenge from group averages.
V. Disproportionate means: The relativized and the absolute view
To expound my third and final challenge, I first need to take a closer look at the disproportionality condition. Characterizations of indirect discrimination contain some such condition. For instance, the Equality Act 2010 definition (see note 2) includes a disproportionality requirement, and in Griggs v. Duke Power the Supreme Court drew upon a proportionality clause to the effect that the exclusion of African-Americans had to be disproportionate in relation to job performance or business necessity.
Disproportionality is a relation between two items. One item—call it the bad item because it is this feature which invites the accusation that the rule or practice is unjustified—is disproportionate relative to another, which we might call the good item, because it can be called on in an effort to show that the rule or practice is justified, e.g., as in “The large amount of force used—a (very) bad item—was disproportionate to the relatively harmless threat thereby averted—a (minor) good item”. To clarify the disproportionality condition we need to say a little more about these good and bad items.
Starting with the former, the first thing to note is that the use of the phrase “legitimate aim” in the Equality Act 2010 can be misleading, in that it suggests that the good item is a certain sought for outcome, not the outcome itself. To see the difference, imagine the Supreme Court had instead found that Duke Power’s high school requirement did indeed represent a business necessity, but also that the company operated this requirement neither with the aim of excluding African-Americans, nor in an effort to maximize business, but for some other reason that was legitimate. For example, the aim was to promote workplace harmony (which was not a business necessity) and the company believed, falsely, that a recruitment process ensuring that all members of senior staff had a high school degree would be one way of achieving this. Here there is no disproportionality, even though the company does not impose the high school requirement out of a concern for business necessity. What matters is that the requirement constitutes a business necessity. More generally, what matters is that there is some consequence (bankruptcy) of not applying the rule (or practice or policy) that justifies it, not whether the avoidance of this consequence is what motivates the agent whose decisions are being assessed for indirect discrimination.
The next question that arises in relation to the good item concerns the nature of the relevant consequences—i.e. the currency of disadvantage. In Griggs v. Duke Power the consequences were couched in economic terms for the company in question. Impact on business was the key consideration. From a legal point of view, this narrow focus might make good sense. The assessment of the broader societal effects of a particular rule is difficult and, hence, to make law sensitive to such effects will make it hard for companies to know if they have infringed indirect discrimination laws. However, from a moral point of view it makes little sense to disregard these less easily quantified effects. For instance, a given admissions test may result in universities doing less well on narrow, university-related parameters (e.g. research output, donations, proportion of students graduating). At the same time, the use of this test rather than an alternative might generate much greater benefits for society generally (e.g. in terms of society being more tolerant and harmonious, and culturally and economically vibrant). In these circumstances, it would seem that, if we want our definition of indirect discrimination to include a disproportionality condition, these broader and beneficial consequences really ought to figure in the disproportionality at issue. Certainly, if the fact that a rule is indirectly discriminatory is a prima facie reason for thinking it is unjust, we should be willing to examine the proportionality of societal effects. Admittedly, doing so may raise more questions than it answers, because now we will now face tricky questions about how to assess a much broader range of consequences; there are many different suggestions as to what makes such consequences good, and as to how they should be weighed against one another. But these questions are not tied specifically to indirect discrimination. They are tied up with much more general issues in moral philosophy. Having flagged them, I will move on.
Let us now turn to the other of the two items in the disproportionality condition: the bad item. Two views here merit examination. The first is that a group is disadvantaged by a rule if, and only if, the inequality between this group and groups with which it is to be compared is greater with the rule than it would be in some relevant alternative situation without it. The second is that a group is disadvantaged by a rule if, and only if, this group would have been better off in some relevant alternative situation without it. Let us call the first view the group-relative (or simply relativized) view, and the second the absolute view.
To see the difference, consider a company that has a choice between two hiring policies. One involves hiring on the basis of qualifications only. The other involves hiring on the basis of qualifications on condition that the group of appointees faithfully reflects the make-up of society as a whole in respect of the protected groups (Spanish-speaking as opposed to English-speaking people, let us say, and suppose that these two groups have the same number of members). It turns out that the second policy results in the company not always hiring best-qualified applicants. This means the company will do less well commercially and end up hiring fewer people. (It is often argued that representational aims improve the competitiveness of a company. I want to steer clear of this empirical issue to address the normative issue of whether indirect discrimination is unjust if it reduces competitiveness in a certain way.) Moreover, in fact, the company will hire more people from any of the protected groups if it always hires the best qualified people than it would if it were to apply the second hiring policy. The situation is as follows:
|Number of English-speaking people hired||Number of Spanish-speaking people hired||Percentage of those hired who are women|
|Hiring policy 1||400||200||Approx. 33%|
|Hiring policy 2||180||180||50%|
Suppose, finally, that we do not have to worry about consequences like objectively demeaning messages, e.g., it is not the case that severe underrepresentation of one group will objectively signify that members of the underrepresented group are inferior and deserve less concern and respect than others (Hellman 2008). On the relative view of the disproportionality condition, the first hiring policy may well be indirectly discriminatory, but the second policy is not so. On the absolute view, the first hiring policy is not indirectly discriminatory, while the second policy is. Indeed it might qualify as a policy that indirectly discriminates against English- and Spanish-speaking people. This implication is strikingly revisionist. It illustrates the general idea that, in principle, inequality is capable of making members of the worse off group better off than they would be under equality. John Rawls appealed to this general idea in defending his renowned “difference principle” of justice. The principle says that, subject to certain constraints, a just society is one in which the worst off in society are as well off as possible. From this it follows that inequalities are tolerable when, and to the extent that, they are required to make the worst off better off (Rawls 1971, pp. 302-303).
Interestingly, the general idea has gone largely unnoticed in discussions of indirect discrimination. Most who work in this area simply assumes a relative view of disadvantage. Thus it is common to find writers inferring, from the underrepresentation of a group, that this group is (probably) being subjected to indirect, if not direct, discrimination (Craig 2007, p.122). Because this inference is clearly invalid on the absolute view, one charitable interpretation of the views of those who make this inference is that they are wedded to the relativized view of disadvantage.
VI. The levelling down objection and indirect discrimination
How does the difference between the relativized and absolute view of disadvantage bear on the claim that indirect discrimination is unjust? In answering this question, I want to bring in what is usually referred to as the “levelling down” objection to egalitarianism—an objection occupying a prominent place in recent discussions of distributive justice, but which has so far not drawn attention in discussions of discrimination. Suppose we subscribe to the following strict egalitarian view: it is “bad—unjust and unfair—” if some people are worse off than others (Temkin 1993, p.13; Parfit 1998, p.3). Apparently, this view implies that a situation in which half the population is at 150 units of whatever is the currency of justice (welfare, resources etc.) and the other half is at 120 is unjust compared to one in which everyone is at 100. On the strict egalitarian view, the second situation, in which everyone is worse off, seems to be in one way better, because less unjust, than the first, in which everyone is better off. It is in one way better because it is better in terms of justice. Yet, as Derek Parfit has argued, this looks implausible. How can one situation be in any way better, e.g. in terms of the justice of distribution, than another in any respect if it is in no respect better for anyone, Parfit asks? (Parfit 1998, p.3). Many have taken this question to lay down a powerful challenge to egalitarianism.10 Moreover, it is even more powerful because there is an alternative to what we have called strict egalitarianism which arguably possesses many of the attractions of that view yet appears to provide an answer to the levelling down objection: “prioritarianism” (See, however, Voorhoeve and Otsuka 2009). The defining idea of prioritarianism is that an equal sized benefit which accrues to a person who is better off on some absolute scale of well-being has less moral value than a benefit that accrues to a person who is worse off on such a scale of well-being. If benefits can be redistributed and the redistribution will not affect the overall sum of benefits, an equal distribution is best, according to prioritarianism. But the levelling down objection has no purchase. Benefits to people, however well off they are, have positive moral value, but that value decreases the better off these people are. The upshot is that a situation where some are worse off and none is better off can never be better in any respect than one in which some are better off and none is worse off.
Let us return now to the conception of indirect discrimination on which we take disproportionality to involve the imposition of relativized disadvantages on the discriminatee. This conception is vulnerable to a challenge similar to the levelling down objection. To see this, suppose that where indirect discrimination occurs the members of one ethnic group will end up with 150 and members of another group 120, and that where it is eliminated everyone ends up with 100. We can now see that if strict egalitarianism is vulnerable to the levelling down objection, the view that indirect discrimination is bad because it is unjust is vulnerable to something very similar. How can indirect discrimination be bad in any respect, e.g. in terms of justice, one might ask, when it is bad in no respect for anyone? This is the levelling down objection to the view that indirect discrimination as such in unjust.
I now want to rebut three critical responses to the levelling down objection to the injustice of indirect discrimination presented in the previous section, although ultimately I will concede that the levelling-down challenge is not decisive. First, then, it might be suggested that if the present challenge is sound, a similar one can be mounted to direct discrimination. However, direct discrimination is indisputably unjust. Hence, the present challenge must contain an error. This response is problematic. Direct and indirect discrimination differ. Assuming that a purely outcome-focused account of fairness is false, the latter is fair in form (recall the formulation in Griggs v. Duke Power), the former is not. Indirect discrimination, if it is unjust, is unjust in virtue of the disadvantages it involves for certain groups (Cf. (Cavanagh 2003, p.199; Alexander Forthcoming). If a company rejects African-Americans on grounds of race, it treats them unfairly and arguably this violates an agent-relative, deontological constraint.11 However, when a company applies a certain test in a way that is indirectly discriminatory the application of the test in itself is not unfair—if it were, the case would probably involve direct discrimination instead. It is only where, in the circumstances, the application of a test disadvantages members of a certain group that injustice is perpetrated.
The second challenge says that the levelling down objection to indirect discrimination is irrelevant, because, as a matter of fact, it never happens that no one is better off in the absence of indirect discrimination and some are even better off. My response to this challenge has three parts. (a) Even if the empirical basis of the challenge is true, this does not render the levelling down objection irrelevant to my question about indirect discrimination. My question is whether indirect discrimination as such is unjust, and to explore this question we need to consider hypothetical cases as well as actual and likely ones. (b) If we ask a different question—namely, one about what we, as political agents trying to bring into being a world that is more just, should be focusing on—the factual assumption is relevant. If, as a matter of fact, the discriminatees in cases of indirect discrimination would be better off if we eliminated that discrimination, we have some reason to do the latter, and this remains so even if there are counterfactual circumstances where doing so would not benefit and perhaps even harm indirect discriminatees. (c) So far I have granted the objector the factual assumption that in all cases of indirect discrimination members of groups suffering it would be better off in its absence. I do not want to claim that this is false (but recall my remark about African-Americans, and Duke Power employees with high school degrees). However, I would point out that it is a very strong claim, and that backing it up with evidence is a daunting task. Moreover, as the debate about affirmative action shows, it is far from uncontroversial that eliminating indirect discrimination always benefits the discriminatee. Thus it has been claimed that a demeaning message is sent when the criteria of assessment are adjusted to favour otherwise underrepresented groups, and that in some cases the resulting message-related costs to such groups of not simply hiring or admitting applicants on a straightforwardly meritocratic basis are too great (Strauss 1995; Adarand v. Pena 1995, p.241; See, however, Bowen and Bok 1998).
Third, in setting out the levelling down objection to indirect discrimination I imagined that the elimination of indirect discrimination would not be better in any way for the discriminatee. However, the disadvantaging of groups involved in indirect discrimination does symbolic harm that ceases to be done when the discrimination is prevented. Hence, it might be suggested that eliminating indirect discrimination is always better in one respect: it eradicates symbolic harm. It is true that some forms of indirect discrimination are symbolically loaded and clearly do affront, or are seen as an affront to, the affected groups. However, this is not true of all forms. The tests used to recruit Navy Seals indirectly discriminate against elderly people, yet they are not generally thought to harm them symbolically. And we can certainly imagine other cases where indirect discrimination would involve no (effective) symbolic harming—e.g. because members of disadvantaged groups remain unaware that they are being disadvantaged by the relevant rules or practices. This shows that indirect discrimination as such does not cause symbolic damage. Finally, even in cases where symbolic harm is involved, the harming might be outweighed, morally speaking, by other kinds of harm that would be done if the indirect discrimination were eliminated. It may so happen, for example, that lowering meritocratic standards would harm all of us, and visit harm on the discriminatees that outweighs the benefit they would enjoy when shielded from symbolic harm.
Since none of the three challenges above is convincing, I tentatively suggest that if the levelling down objection defeats strict egalitarianism, it defeats the view that indirect discrimination is unjust. Like egalitarianism, concern about indirect discrimination arises from uneasiness at the relative positions of different groups. This opens the door to the levelling down objection, because one can always imagine the relative positions being adjusted in a way that leaves everyone worse off in absolute terms than they were before the adjustment. As this formulation indicates, the feature of a view of distributive justice that makes it vulnerable to the levelling down objection is not that it claims that justice is equality, but that it claims that justice consists in a certain relation between people’s distributive positions. A view according to which justice requires that no one is (or indeed one that requires that some are) significantly worse off than others is also vulnerable to the levelling down objection. For simplicity I disregard this broader scope of the levelling down objection and simply focus on strict egalitarianism (See Lippert- Rasmussen Forthcoming).
Does the levelling down objection amount to a knockdown argument against the view that indirect discrimination is unjust? I am not sure. Strict egalitarians have developed responses to the levelling down objection which, suitably revised, can be deployed in a rearguard action here. Some have pointed out that values other than equality imply that one outcome can better than another, even if it is better for no one in any respect. In a retributivist perspective on criminal justice, for instance, a world in which criminals are justly punished might be assessed as better than one in which they are not, even if this is better for no one because punishment has no deterrent effect. Hence, if the Slogan that “One situation cannot be worse (or better) than another in any respect if there is no one for whom it is worse (or better) in any respect” (Temkin 1993, p.248) obliges us to reject a wide range of values other than equality, perhaps the intuitive cost of rejecting it is lower than the intuitive cost of rejecting equality, desert and all the other values that offend against the Slogan (Temkin 1993, p.261).
In a separate move, it has been argued that some of those who reject egalitarianism in response to the levelling down objection are not really in a position to do so (Persson 2008). Consider prioritarianism. On this view, if we transfer one unit of well-being from a well off person to a badly off person this will result in an increase in moral value. But where does this increase come from, one might ask? Ex hypothesi, the decrease in well-being experienced by the source is exactly as great as the increase in well-being experienced by the recipient of the well-being. Accordingly, the value the transfer brings into existence seems to be unconnected to the sums of well-being. This suggests that prioritarianism, like egalitarianism, is committed to the idea that values are not tied to well-being for individuals. Since prioritarianism is commonly adopted by those who press the levelling down objection, this reversal of the attack has considerable bite.
Finally, some egalitarians take a bullish stance: they insist that, because it follows straightforwardly from strict equality that a state in which everyone is worse, but equally well off, is in one respect—though not all things considered—better than one in which everyone is better off, though unequally so, this implication is something they were aware they were committed to all along. Accordingly, the levelling down objection cannot play the dialectical role of an objection—it does not point to an implausible implication to which egalitarians are committed and of which (until the alleged objection was presented to them) they were unaware. Ironically, in view of Parfit’s formulation of the levelling down objection, it is probable that more egalitarians now will take this attitude than would have done so 30 years ago.12
There is a huge literature on the levelling down objection, and my aim here is not to argue that it refutes the claim that indirect discrimination can be unjust (see Holtug 2011, pp. 181- 201). My arguments in Section III and Section IV suffice to support this conclusion. My main aim here is to argue that indirect discrimination is unjust only if a strict egalitarian view of justice is correct, and thus that the levelling down objection fails. Even if the result of the discussion is limited in this way, it is very significant. Many people who are not committed to strict egalitarianism think that discrimination, including indirect discrimination, is unjust. Indeed, one hallmark of contemporary liberal opposition to discrimination is the assumption that one can be opposed to discrimination without committing oneself to any form of strict egalitarianism. If the argument of this section is sound, this option is unavailable, at least in the case of indirect discrimination. Strict egalitarianism of a certain sort—i.e. one that focuses on socially salient groups—is tied to the view that indirect discrimination is unjust! Since many would not want to tie them together in this way, my claim that they stand and fall together forms my third challenge to the view that indirect discrimination as such is unjust.
If the reasoning behind the local-global disadvantage dilemma, the challenge from group averages, and the levelling down objection applied to indirect discrimination is sound, indirect discrimination is not necessarily unjust. Because I am not certain that the levelling down objection is successful, my own basis for asserting the main claim of this article derives from the first two reasons only. I put forward the last objection, in its non-conditional form, in an ad hominem way.
Some might find the claim that indirect discrimination is not necessarily unjust discomforting. For one thing, they might worry that anyone who is persuaded by them will have to approve the legalization of indirect discrimination and (more generally) stop worrying about it. Let us briefly consider whether these worries are warranted.
First, it does not follow from the fact that something is not unjust as such that it is not often unjust. Contracts between employers and employees are not unjust as such, but many are unjust all the same—e.g. because they involve exploitation of the vulnerability of the employee by the employer. This means that the arguments in this article are entirely compatible with the view (which I am neither affirming, nor denying, here) that many forms of indirect discrimination should be made unlawful, because they are unjust. Moreover, to the extent that one allows that something might be unlawful, not because it is unjust, but because its presence often indicates injustice elsewhere, one could also, consistently with what I have argued, hold that indirect discrimination should be unlawful.13
Second, even if indirect discrimination is neither unjust, nor even often unjust (or even sometimes unjust), we have to remember that justice is not the only moral value, and that other values might speak against indirect discrimination. For instance, the French Revolution famously acclaimed fraternity as well as liberty and equality. Arguably, fraternity is hard to realize in a society where some groups are seriously underrepresented in the most prestigious and well paid job categories (Anderson 2010, pp. 89–111; Cohen 2009, pp. 27-34). So even if such underrepresentation is not unjust, it might still be morally indefensible, all things considered, not to eliminate the indirect discrimination that brings about such underrepresentation.
Assessment of the strength of my arguments should, therefore, proceed independently of the worries mentioned above. In light of the remarks made above, however, another worry might arise. The question would be: if the view that indirect discrimination is not unjust is compatible with its being the case that indirect discrimination ought, morally speaking, to be unlawful, and with measures that are normally taken to counteract its effect, does this article have any significant practical implications at all? I believe the answer is yes, and that this article has two very significant practical implications. The first is that we cannot infer from the fact that a certain group is underrepresented that it is being treated unjustly, just as we cannot infer from the fact that it is overrepresented—witness, my example in the next paragraph—that it enjoys discrimination in its favour. This is significant, because these inferences of this sort are often made (and often criticized).
The second significant implication is that we will have to think more about what it is that makes cases involving indirect discrimination just or in other ways morally wrong. Take admission rules at Ivy League universities that result in the numerical “overrepresentation” of Asian-Americans. There is an obvious sense in which such rules disadvantage non-Asian-Americans, yet we would not consider this unjust, indirect discrimination. But then why are we inclined to infer this, when the underrepresented group is African-American instead? Enquiries such as the present one force us to try to identify the morally relevant difference. Moreover, they suggest that there are such differences, but that they are not necessarily best thought of in discrimination-related terms. Also, the present enquiry forces us to think hard about the relationship between strict egalitarianism and the injustice of indirect discrimination. In these two ways, and despite the nuances mentioned above, the present article does have significant practical implications. Various forms of affirmative action might well be morally justified, but the present line of argument suggests, surprisingly, that such justification may have little to do with the need to eliminate the injustice of indirect discrimination.
Acknowledgments: Previous versions of this paper were presented at University of Copenhagen, Université catholique de Louvain, the Swedish Congress of Philosophy, and University of Roskilde. I wish to thank Simon Caney, Åsa Carlsson, Axel Gosseries, Rune Klingenberg Hansen, Deborah Hellman, Nils Holtug, Adam Hosein, Magnus Jedenheim-Edling, Sandra Lindgren, Sune Lægaard, Søren Flinch Midtgaard, Thomas Søbirk Petersen, Christian Rostbøll, Ruth Rubio-Marin, Jesper Ryberg, Jens Damgaard Thaysen, and Frej Klem Thomsen for helpful comments. I am also grateful to Larry Alexander for an insightful written response.
Larry Alexander, “What Makes Wrongful Discrimination Wrongful? Biases, Preferences, Stereotypes, and Proxies” University of Pennsylvania Law Review 141.1 (1992), 149–219.
——— “Disparate Impact: Fairness or Efficiency?”, San Diego Law Review 50.1 (forthcoming).
Elizabeth S. Anderson, The Imperative of Integration (Princeton, NJ: Princeton University Press, 2010).
Richard Arneson, “Against ‘Complex’ Equality”, in David Miller and Michael Walzer (eds.), Pluralism, Justice, and Equality (Oxford: Oxford University Press, 1995), pp. 226–252.
William G. Bowen and Derek Bok, The Shape of the River: Long-Term Consequences of Considering Race in College and University Admissions (Princeton, NJ: Princeton University Press, 1998).
John Broome, Weighing Goods (Oxford: Basil Blackwell 1991).
Matt Cavanagh, Against Equality of Opportunity (Oxford: Oxford University Press, 2003).
G. A. Cohen, Why Not Socialism? (NJ: Princeton University Press, 2009).
——— On The Currency of Egalitarian Justice and Other Essays in Political Philosophy (Princeton, NJ: Princeton University Press, 2011).
Michael Connolly, Discrimination Law (London: Sweet and Maxwell, 2011).
Kimberlé Crenshaw, “Demarginalizing the Intersection of Race and Sex: A Black Feminist Critique of Antidiscrimination Doctrine, Feminist Theory and Antiracist Politics,” in Anne Phillips (ed.), Feminism and Politics (New York: Oxford University Press, 1998), pp. 314–343.
DH v. Czech Republic, (Application no. 57325/00) 47 EHRR 3 (2008).
Oran Doyle, Oran “Direct Discrimination, Indirect Discrimination and Autonomy”, Oxford Journal of Legal Studies 27.3 (2007), pp. 537–553.
Sandra Fredman Discrimination Law 2. ed. (Oxford: Oxford University Press, 2011).
John Gardner, “Discrimination as Injustice”, Oxford Journal of Legal Studies 16.3 (1996), pp. 353–367
Lena Halldenius, “Dissecting ‘Discrimination’”, Cambridge Quarterly of Healthcare Ethics 14.4 (2005), 455–463.
Patricia Hartnett, “Nature or Nurture, Lifestyle or Fate: Employment Discrimination Against Obese Workers”, Rutgers Law Journal 24.3 (1992—1993), pp. 807–845.
Deborah Hellman, When is Discrimination Wrong? (Cambridge: Harvard University Press, 2008).
Nils Holtug, Persons, Interests, and Justice (Oxford: Oxford University Press, 2011), pp. 181–201.
Nils Holtug and Kasper Lippert-Rasmussen, “Introduction”, in Holtug and Lippert-Rasmussen (eds.) Egalitarianism: New Essays on the Nature and Value of Equality (Oxford: Oxford University Press, 2007), pp. 6-7.
Frances Kamm, Intricate Ethics (Oxford: Oxford University Press, 2007), pp. 24, 170-173, on the Principle of Secondary Permissibility.
Klemens Kappel and Peter Sandøe, “QALYs, Age and Fairness”, Bioethics 6.4 (1992), pp. 297–316.
Kasper Lippert-Rasmussen, “The Insignificance of the Distinction between Telic and Deontic Egalitarianism” in Holtug and Lippert-Rasmussen (eds.) Egalitarianism: New Essays on the Nature and Value of Equality (Oxford: Oxford University Press, 2007), pp.101-124.
——— Born Free and Equal? A Philosophical Inquiry Into the Nature of Discrimination (New York: Oxford University Press, 2013), pp. 38-40.
——— “Distributive Justice and Discrimination”, in Serena Olsaretti (ed.), Oxford Handbook to Distributive Justice (Oxford: Oxford University Press, forthcoming).
Andrew Mason, “Egalitarianism and the Levelling Down Objection”, Analysis 61.3 (2001), pp. 246–254.
James W. Nickel, “Discrimination”, in Edward Craig and Edward Craig (eds.) Concise Routledge Encyclopedia of Philosophy (London: Routledge, 2000), p. 214.
Derek Parfit, “Equality and Priority,” in Andrew Mason (ed.) Ideals of Equality (Oxford: Blackwell Publishers, 1998), pp. 1 – 20.
Ingmar Persson, “Why Levelling Down Could be Worse for Prioritarianism than for Egalitarianism”, Ethical Theory and Moral Practice 11.3 (2008), pp. 295–303.
John Rawls, A Theory of Justice (Oxford: Oxford University Press, 1971).
Ricci v. DeStefano (Nos. 07–1428 and 08–328) 530 F. 3d 87 (2009), .
Frederic Schauer, Profiles, Probabilities, and Stereotypes, (Cambridge, MA: Harvard University Press, 2003), pp. 224–250.
Michael Selmi, “Was Disparate Impact Theory a Mistake?”, UCLA Law Review 53 (2006), pp. 701–782.
Shanagan v. UK (Application no. 37715/97) 2001; Council Directive 2000/43/EC of 29 June 2000, article 2(b), http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=CELEX:32000L0043:en:HTML [accessed January 5, 2014].
Peter Singer, “Is Racial Discrimination Arbitrary?”, Philosophia 8.2–3 (1978), 185–203.
Charles A. Sullivan, “The World Turned Upside Down? Disparate Impact Claims by White Males”, Northwestern University Law Review 98.4 (2004), 1505-1565.
David A. Strauss, “Affirmative Action and the Public Interest”, The Supreme Court Review, 1995 (1995), pp. 1–43.
Larry S. Temkin, Inequality (Oxford: Clarendon Press, 1993).
Alex Voorhoeve and Michael Otsuka, “Why it Matters that Some are Worse Off than Others: An Argument against the Priority View”, Philosophy & Public Affairs, 37.2 (2009), pp. 171–199.
Michael Walzer, Spheres of Justice (Oxford: Basil Blackwell, 1983).
Alan Wertheimer, “Jobs, Qualifications, and Preferences”, Ethics, 94.1 (1983), 99–112.
Jonathan Wolff and Avner De-Shalit, Disadvantage (New York: Oxford University Press, 2007).
1. For a similar claim in relation to so-called reaction qualifications in general, see (Wertheimer 1983, p.101; Alexander 1992, , pp. 173–176)
2. A similar legal stance is represented by Britain’s Equality Act 2010, which prohibits direct as well as indirect discrimination in relation to certain “protected characteristics”: “age; disability; gender reassignment; marriage and civil partnership; race; religion or belief; sex; sexual orientation.” The Act states that “(1) A person (A) discriminates against another (B) if A applies to B a provision, criterion or practice which is discriminatory in relation to a relevant protected characteristic of B’s. (2) For the purposes of subsection (1), a provision, criterion or practice is discriminatory in relation to a relevant protected characteristic of B’s if—(a) A applies, or would apply, it to persons with whom B does not share the characteristic, (b) it puts, or would put, persons with whom B shares the characteristic at a particular disadvantage when compared with persons with whom B does not share it, (c) it puts, or would put, B at that disadvantage, and (d) A cannot show it to be a proportionate means of achieving a legitimate aim.” (Equality Act 2010), cf. (Connolly 2011, pp. 55–77).
3. For discrimination against obese people, see (Harnett 1992-1993). For income discrimination, see (Lippert-Rasmussen 2013).
4. I take it that if something is unfair it involves a violation of comparative justice.
5. For some readers it may be helpful to note that I am exploring whether indirect discrimination is pro tanto unjust.
6. Altman’s definition implies that it is only socially salient groups that can be subjected to indirect discrimination. I omit this part of his definition, because, as noted in Section I in relation to the issue of the nature of protected groups, my focus is on issues that pertain specifically to indirect discrimination, as opposed to discrimination in general; but see (Lippert-Rasmussen, 2013, chapter 1). Sometimes people use a moralized concept of indirect discrimination such that if something is indirect discrimination, it is by definition unjust (or morally unjustified). I set aside this concept here. The discussion I present can be read as showing that much of what people who employ the moralized concept identify as indirect discrimination does not fall under their concept.
7. For a reply defending the view that it is the distribution of global benefits and harms that matters, see (Arneson 1995)
8. This example brings out the core issue of intersectionality and discrimination: that, at one and the same time, individuals might be discriminated against and in favor of in many different capacities; see (Crenshaw 1998)
9. For an account of the difference between political advocacy and political philosophy, see (Cohen 2011, pp. 225–235.)
10. Admittedly, Parfit (1998, pp. 6-7) seems to suggest that a certain form of egalitarianism—deontic egalitarianism according to which it is the way in which inequality is produced and not the unequal outcome in itself that is unjust—is not vulnerable to the levelling down objection. Elsewhere I have argued that deontic egalitarianism is so vulnerable if telic egalitarianism is (Lippert- Rasmussen 2007)). In any case, the very idea behind indirect discrimination is that its injustice lies in the unequal outcome it generates, not in the indirectly discriminatory acts themselves, which after all are “fair in form”. Accordingly, I do not see how an objection to indirect discrimination could derive from deontic egalitarianism. More generally, I do not see which agent-relative restriction pertaining the “act itself”, so to speak, that someone who indirectly discriminates can plausibly be said to violate. For instance, I do not think it is plausible that there is a deontological restriction against indirect discrimination where indirect discrimination makes people better off in the way explored in Section III, see (Kamm 2007, pp. 24, 170-173)
11. For some doubts about the view that the levelling down objection, mutatis mutandis, does not challenge deontological views of justice too, see (Lippert-Rasmussen 2007)
12. Another response to the levelling down objection is to hold that equality is non-instrumentally valuable, but that it is so only on condition that it benefits someone: see (Mason 2001) Yet another response is that, necessarily, unjust inequality is bad for worse off people: see (Broome 1991, p. 165)
13. See the discussion in (Schauer 2003), of presumed offenses. For instance, in Bentham’s days it was a presumed offence to alter a ship’s officially registered name. Obviously, to do so is not an offence in itself, but one can presume that often such a change of name is motivated by a malign reason, i.e. to disguise that the ship has been stolen from its rightful owner.