• 22 June 2009

Passive Discrimination

Jonah Gelbach & Lesley Wexler & Jonathan Klick

Posted in , ,

In this Editorial, we present a distinct mechanism of employer discrimination largely ignored by scholars and regulators alike.  What we term “passive discrimination” involves an employer’s use of wage and benefits packages that exploit observed, systematic group-level preference heterogeneity in order to induce worker sorting such that members of a disfavored group view the job opportunity as being less attractive than do members of other groups. A companion to this Editorial, which lays out the formal model, can be found on the Legal Workshop website here.

By way of illustration, imagine that individuals from two groups, Deltas and Omegas, comprise the labor pool from which employees may be hired.  While their work productivity is drawn from the same distribution, a given employer dislikes Omegas for reasons unrelated to their job qualifications.  Because Omegas have suffered discrimination historically, legislation explicitly protects Omegas from employment discrimination.  Further, while Deltas and Omegas have similar reservation wages,1 Deltas, on average, more highly value some nontransferable good that the employer can procure (or produce) at a cost equal to or lower than the Deltas’ average valuation of the good.  To make the illustration more concrete, assume the employer is a brewery and offers free beer at lunchtime.

While the employer prefers to hire only Deltas, federal legislation limits the employer’s ability to do so. Yet we contend that employers may advertise the job broadly and make hiring decisions in a seemingly nondiscriminatory fashion, avoiding lawsuits, and still achieve an ultimate workforce that is predominantly (if not exclusively) composed of Deltas.  Specifically, if the employer offers a compensation package composed of a submarket wage as well as access to free lunchtime beer, Omegas will find such a job unattractive, while Deltas will still gladly accept the job offers.  Ultimately, according to the firm, due to no misconduct on its part, Omegas lacked interest in working for the brewery.

While scholars have touched on this phenomenon,2 no one has examined these employment practices in a systematic way.3 Such discrimination is presented as an ancillary effect of employment policies or conditions. Yet recent class actions suggest that basic animus- or stereotype-driven discrimination is still quite prevalent. So it should be unsurprising that litigation-savvy employers might deliberately craft compensation structures to exclude certain types of workers.

In this online Article, we provide an illustration of group-level preference heterogeneity that could generate passive discrimination in an occupational setting. We next discuss how current antidiscrimination law applies and conclude with a brief discussion of how we might better address passive discrimination.


In this Part, we first describe the conditions necessary to create the segregated equilibria described above. We then provide an example where these conditions may apply. In the case of intentional passive discrimination, the employer must be able to identify a good that satisfies two conditions: (1) the disfavored group values the good on average less than the average valuation placed on the good by other groups of potential employees, and the employer knows this; and (2) the good is nontransferable. A third condition under which the employer can ensure that she attracts the favored type of employee is satisfied when the employer can provide the good for a cost below that paid by workers outside of the employment relationship.

While we have focused on intentional passive discrimination, the phenomenon could also arise as an accidental byproduct.  For example, a brewery whose owners are indifferent regarding employing Deltas and Omegas might still offer free beer during lunch because it believes doing so promotes product knowledge.  In such a case, even if the brewery starts out offering this benefit plus a market wage, Deltas who do not secure positions will offer to work for a lower wage given their valuation of the free beer.  Eventually, this group valuation will lead to a workforce composed of Deltas, to the exclusion of Omegas.

One example of potentially discriminatory screening practices relates to subjective discount rates and pay structures that include a deferred compensation component, such as a pension.4 To begin with, a person’s subjective discount rate captures her willingness to delay current consumption for the prospect of increased future consumption.  While everyone exhibits some positive subjective discount rate, individual-to-individual heterogeneity exists in those subjective discount rates.5

While individual-level heterogeneity is prevalent, labor economists have noted that systematic differences in individual discount rates across racial groups may also exist.6 A natural experiment provides the most interesting supporting evidence in the finding of large inter-race heterogeneity in discount rates.7 Specifically, conditional on a large number of other effects,8 black military enlistees and officers exhibited significantly higher subjective discount rates than other minorities and whites.9 This result is robust with blacks exhibiting, on average, discount rates on the order of five to nine times as great as whites.

Assuming this empirical regularity holds,10 an employer wishing to passively exclude blacks could offer a low current wage coupled with generous deferred compensation benefits, such as a large pension. Such a package would attract individuals with relatively low subjective discount rates and repel those with higher subjective discount rates. Note further that this could also represent a situation in which a nondiscriminatory employer might engage in unintentional passive discrimination, as employers may possess numerous other reasons to include a generous retirement component in its compensation package.

Applying Title VII

In determining the legality of passive discrimination, we begin with Title VII. Congress enacted this statute “to assure equality of employment opportunities and to eliminate those discriminatory practices and devices which have fostered [ ] stratified job environments to the disadvantage of minority [or other protected] citizens.”11 Congress did not ban all forms of workplace segregation, but rather forecasted that integration would be a beneficial byproduct of ending active discrimination. Thus, Title VII makes it an unlawful employment practice for an employer to “fail or refuse to hire . . . or otherwise to discriminate against any individual with respect to his compensation, terms, conditions, or privileges of employment, because of such individual’s race, color, religion, sex or national origin.”12

A.     Disparate Treatment

Title VII jurisprudence allows plaintiffs to choose among disparate treatment and disparate impact claims. Under disparate treatment, an employer treats some individuals worse than others because of a protected characteristic. “Proof of discriminatory motive is critical, although it can . . . be inferred from the mere fact of differences in treatment.”13 In individual disparate treatment claims, each plaintiff must prove that she was treated less favorably than others similarly situated and that this disparate treatment was “because of” the plaintiff’s race, color, sex, national origin, or religion. The plaintiff must provide either direct or circumstantial evidence to demonstrate the employer’s discriminatory intent. While these claims may evidence a concern about segregation, courts link this concern to the elimination of discriminatory practices which deprive individuals of employment opportunities.14

In pattern or practice cases, another type of disparate treatment claim, the plaintiff can satisfy the prima facie case with “statistical evidence demonstrating substantial disparities in the application of employment actions as to minorities and the unprotected group.”15 These plaintiffs need not present individual victim testimony to support a finding of intentional discrimination—courts may rely purely on evidence of gross statistical disparity,16 which raises an inference of discriminatory intent.

Under disparate treatment claims, Title VII prohibits the employer from using group-based characteristics, preferences, or stereotypes to treat individuals differently even if such stereotypes are largely accurate. In City of Los Angeles, Department of Water and Power v Manhart,17 the Supreme Court ruled that an employer may not deduct more from women’s pay to cover their pensions even though as an actuarial matter, women as a class are likely to draw more pension benefits.18 The Supreme Court reasoned that any individual woman may not draw benefits longer than any individual man, her employers may not condition her pay on her sex.19

Easy disparate treatment cases include those instances in which an employee offers a potential plaintiff a facially different wage than others similarly situated. For example, an employer may not offer Caucasian bus drivers comprehensive health insurance and fail to offer African-American drivers the same insurance package. If, on the other hand, the employer offers all bus drivers lower wages and higher pension benefits than other area employers, individual employees cannot successfully lodge a disparate treatment claim. Even if empirical evidence indicates that, as a group, African-American drivers have a higher discount rate and place a lower value on such pensions, the employer has treated each individual African-American driver the same as all its other drivers. So we contend no individual disparate treatment claim could succeed.

Under pattern-and-practice claims, however, if employers devised a very successful sorting mechanism, plaintiffs might be able to satisfy the prima facie showing of gross statistical disparity. For instance, in International Brotherhood of Teamsters v United States,20 the Court suggested that the complete, or very nearly complete, absence of members of a protected class in a particular job can compel an inference of discrimination.21 So if the high pension, low wage strategy resulted in a workforce with no or very few African-American drivers, employers could face a Title VII problem. Yet plaintiffs rarely prevail in such cases without testimony about individual acts of disparate treatment.22

Only if the employer foolishly allowed the discovery of direct evidence of discriminatory intent would the employer face real difficulty in providing nondiscriminatory explanations for the disparity. Even then, the plaintiffs might not prevail. Such a memorandum would reveal an intent to achieve a segregated workplace, but a plaintiff would still bear the burden of proving the occurrence of discrimination as defined by Title VII. Here, the employer has relied on preferences that tend to be correlated with protected characteristics, but has offered each individual the same package.

Title VII clearly prohibits employers from treating individuals differently because of discriminatory animus or outmoded stereotypes or even seemingly rational group-based stereotypes. Yet, as currently conceived, disparate treatment claims seemingly do not prohibit the employer from using group-based characteristics, preferences, or stereotypes to treat individuals similarly in hopes that such treatment will encourage applicants from disfavored groups to sort themselves out of a job based on their own preferences. As any individual applicant may defy the stereotype and elect into the job, no disparate treatment has occurred even if an employer succeeds in achieving a segregated workplace.

B.     Disparate Impact

Disparate impact claims allow plaintiffs to prevail if they identify a particular employment practice with a significant adverse impact on a protected class, and the defendant fails to demonstrate that the challenged employment practice is “job related  . . .  and consistent with business necessity.”23 If the challenged practice significantly serves the employer’s legitimate employment goals,24 the plaintiff can still prevail if she proves that a less discriminatory alternative employment practice equally serves the defendant’s goals.25

Passive discrimination seemingly fits better under this analysis, as employers devise facially neutral compensation packages that may result in only a few individuals from a disfavored group in their workplace. Yet the Supreme Court has cast doubt as to whether fringe benefits and compensation packages are subject to disparate impact analysis.26 In Manhart, described above, the Court stated in dicta:

Even under Title VII itself—assuming disparate-impact analysis applies to fringe benefits—the male employees would not prevail. Even a completely neutral practice will inevitably have some disproportionate impact on one group or another. [Griggs v Duke Power Co, 401 US 424 (1971)] does not imply, and this Court has never held, that discrimination must always be inferred from such consequences.27

Likely as a result of this language, few cases have grappled with compensation and fringe benefits under disparate impact analysis. In Finnegan v Trans World Airlines, Inc,28 the Seventh Circuit held that across-the-board cuts in fringe benefits were not eligible for disparate impact analysis under the Age Discrimination in Employment Act.29 Judge Richard Posner rejected even a prima facie case of disparate impact for such cuts, as the focus of disparate impact should be on the exclusion of individuals from certain opportunities.30 But passive discrimination does not exclude anyone from an opportunity; it just makes the opportunity less desirable.

C.     Lack of Interest

The so-called “lack of interest defense,”31 available in both disparate treatment32 and disparate impact cases, is particularly relevant to the causation questions raised by passive discrimination. Lack of interest is a nondiscriminatory explanation for statistical disparities,33 and rebuts the plaintiff’s prima facie case.34 Under this “defense,” employers dispute the causal chain by showing that employees’ voluntary choice, rather than a particular employment practice, causes workplace inequality or segregation.35 If the employer succeeds in showing that individual preferences cause a disparate impact, then it need not reach the question of whether the practice is job related and consistent with business necessity.36

This Article suggests that labor market conditions that shape individuals’ interest in particular jobs may include more than the substance of the job, but also the terms, conditions, and privileges of employment, such as the compensation structure. Yet employers and courts seem to view these packages as mostly within the employers’ discretion to design and the employees’ discretion to take or leave. Passive discrimination suggests that courts’ acceptance of such preferences under the lack of interest doctrine may allow employers to use such preferences with impunity.


This Article fits within a larger debate about the appropriate framework from which to address workplace discrimination and segregation. If one were concerned about workplace segregation or the effect that group-based preferences have on individuals, some judicial and legislative actions might be taken.37 Title VII reform provides one obvious approach. Courts could apply disparate impact doctrine to the structure of compensation and to the provision of fringe benefits. Interest groups may push for stand-alone legislation to address particular mechanisms of passive discrimination. Independent legislation may bypass litigation hurdles associated with Title VII if it does not rely on individual claimants. More innovative approaches include standalone legislation, education initiatives, and incentivized employer restructuring through an enhanced range of employee choice in compensation options, which are more fully fleshed out in our full article.dingbat


Copyright © 2009 The University of Chicago Law Review.

Jonah Gelbach is Associate Professor of Economics, University of Arizona; Jonathan Klick is Professor of Law, University of Pennsylvania Law School; and Lesley Wexler is Assistant Professor of Law, Florida State University College of Law.

The following is a companion to this Editorial: Jonah Gelbach, Jonathan Klick & Lesley Wexler, A Formal Model of Passive Discrimination, LEGAL WORKSHOP (U. CHI. L. REV. Aug. 10, 2009).

The following is a Response by Richard Epstein to this series of Editorials: Richard A. Epstein, Protect Us, Lord, from Title VII: A Response to Gelbach, Klick, and Wexler, LEGAL WORKSHOP (U. CHI. L. REV., June 22, 2009).

  1. Reservation wages are the lowest wage at which an individual is willing to accept a job.
  2. See, for example, Laura T. Kessler, The Attachment Gap: Employment Discrimination Law, Women’s Cultural Caregiving, and the Limits of Economic and Liberal Legal Theory, 34 Mich J L Reform 371, 413-14 (2001) (noting the widespread industry norm of long hours and extensive travel disadvantages women who disproportionately tend to be primary caregivers).
  3. Many have discussed practices that implicate the work-life balance, which may operate to screen out many women from particular jobs. We consider strategies such as the use of long hours and high wages, or substantial face time and high wages, to be special cases, as they directly implicate productivity. As we explain in Part II, for purposes of disproving Gary Becker’s theory, our hypotheticals presume all workers are equally productive. Cases that integrate productivity are important and doctrinally interesting, but we focus on the simplest examples in this Article and leave more complex cases for later works.
  4. In our full Article, we also provide examples that deal with sex, national origin, and religion. See Jonah Gelbach, Jonathan Klick, and Lesley Wexler, Passive Discrimination:When Does It Make Sense to Pay Too Little?, 76 U Chi L Rev 801, 822-27 (2008).
  5. As a general rule, if the amount the individual gives up now is represented by PV and the smallest amount the individual is willing to accept in compensation at the end of n periods is represented by FV, then the individual’s subjective discount rate per period is calculated as: i = {(FV/PV) ^ (1/n)} – 1.
  6. This does not mean, of course, that every individual of race Y is likely to exhibit a higher discount rate than every individual of race Z, but rather that the average discount rate among individuals of race Y will sometimes diverge from the average discount rate among individuals of race Z.
  7. See John T. Warner and Saul Pleeter, The Personal Discount Rate: Evidence from Military Downsizing Programs, 91 Am Econ Rev 33, 33-34 (2001).
  8. The various controls included sex, number of dependents, education level, wage level, benefit level, year of decision, age, years of service, geographic region, service branch, IQ score, and specialty controls. See id at 43-49.
  9. Id at table 4 and table 5.
  10. We make no general claim as to the validity of the empirical finding except to note the high quality of the research papers we cite finding this result. Further, we most certainly do not offer an explanation for why subjective discount rate heterogeneity may follow this pattern.
  11. McDonnell Douglas Corp v Green, 411 US 792, 800 (1973).
  12. Title VII of the Civil Rights Act of 1964, 42 USC § 2000e-2(a)(1).
  13. International Brotherhood of Teamsters v United States, 431 US 324, 335 n 15 (1977) (noting that “disparate treatment was the most obvious evil Congress had in mind when it enacted Title VII”).
  14. See, for example, Marion v Slaughter Co, 1999 WL 1267015, *6 (10th Cir) (observing that the existence of a segregated workplace is not per se a violation of Title VII).
  15. EEOC v Sears, Roebuck & Co, 839 F2d 302, 308 (7th Cir 1988) (citation omitted) (explaining that once the plaintiff has satisfied the initial burden, the burden shifts to the “employer to defeat the prima facie showing of a pattern or practice by demonstrating that the { } proof is either inaccurate or insignificant”).
  16. See Hazelwood School District v United States, 433 US 299, 307-08 (1977) (“Where gross statistical disparities can be shown, they alone may in a proper case constitute prima facie proof of a pattern or practice of discrimination.”).
  17. 435 US 702 (1978).
  18. Id at 711. In Manhart, the defendant used mortality tables and its own experience to determine that the cost of a pension for the average retired female would be greater than for the average retired male. The city required female employees to make greater monthly contributions to the pension fund, which reduced the women’s take-home pay. Id at 705.
  19. Id at 708. Notably, the Court rejected the argument that facially equal deductions might impose a disparate impact on men who as a class were less likely to benefit as fully from the pension plan. Id at 708-09.
  20. 431 US 324 (1977).
  21. See id at 339-40 (“{O}ur cases make it unmistakably clear that statistical analyses have served and will continue to serve an important role in cases in which the existence of discrimination is a disputed issue.”) (quotation marks omitted). See also EEOC v Andrew Corp, 1989 WL 32884, *14 (ND Ill).
  22. For a view that statistics alone are not compelling, see Sears, 839 F2d at 360 (Cudahy dissenting) (suggesting that “the EEOC as much as gave the case away by failing to produce any flesh and blood victims of discrimination. Regression statistics by themselves only demonstrate correlations between variables; to move from correlation to causation, there must be some independent theory about the causal relationships of the variables”).
  23. 42 USC § 2000e-2(k)(1)(A)(i).
  24. See, for example, Watson v Fort Worth Bank and Trust, 487 US 977, 998-99 (1988).
  25. 42 USC § 2000e-2(k)(1)(A)(ii) (providing that “{a}n unlawful employment practice based on disparate impact is established . . . {if} the complaining party makes the demonstration described . . . with respect to an alternative employment practice and the respondent refuses to adopt such alternative employment practice”). See, for example, Albermarle Paper Co v Moody, 422 US 405, 425 (1975).
  26. Fringe benefits cases under disparate impact have dealt with the exclusion of particular benefits, such as contraceptives or fertility treatments, rather than the decision to provide compensation in the form of fringe benefits. See Douglas Laycock, Continuing Violations, Disparate Impact in Compensation, and Other Title VII Issues, 49 L & Contemp Probs 53, 54 (1986) (contending that the logical implication of Manhart is that “{t}here is no disparate impact liability in sex discrimination in compensation cases”). Though another possible reading might merely indicate that disparate impact analysis is only available to minorities and not men, or that where “disparate impact to one group results from avoiding disparate treatment of another, the practice is justified by a business necessity.” Charles A. Sullivan, The World Turned Upside Down?: Disparate Impact Claims by White Males, 98 Nw U L Rev 1505, 1530 (2004).
  27. 435 US at 710 n 20 (citations omitted).
  28. 967 F2d 1161 (7th Cir 1992).
  29. Id at 1163. The court noted that allowing such cuts to be eligible for disparate impact analysis “would mean that every time an employer made an across-the-board cut in wages or benefits he {would be} prima facie violating the age discrimination law. Practices so tenuously related to discrimination, so remote from the objectives of civil rights law, do not reach the prima facie threshold.” Id at 1165.
  30. Id at 1164-65. Judge Posner explained: “The concept of disparate impact was developed for the purpose of identifying situations where, through inertia or insensitivity, companies were following policies that gratuitously-needlessly-although not necessarily deliberately, excluded black or female workers from equal employment opportunities.” Id at 1164 (emphasis added). Finnegan is distinguishable from most of the passive discrimination we discuss, as it was both unintentional and was a response to economic pressures, rather than the original design of the compensation package.
  31. Many courts refer to the “lack of interest defense,” though defendants deploy this argument not as a formal affirmative defense, but as a way to rebut the inference of causation that is raised by statistical disparity.
  32. Courts have recognized the “lack of interest defense” as available even in pattern and practice cases that rely on the inexorable zero. See EEOC v O & G Spring and Wire Forms Specialty Co, 38 F3d 872, 874 n 1 (7th Cir 1994).
  33. See, for example, Sears, 839 F2d at 313 (allowing defendant to use a variety of evidence to demonstrate that women are less interested in commission sales positions than men).
  34. See Teamsters, 431 US at 360 n 46.
  35. For example, in Sears, the EEOC claimed that Sears “engaged in a nationwide pattern or practice of discrimination against women . . . by failing to hire and promote females into commission sales positions on the same basis as males.” 839 F2d at 307. Although the EEOC presented statistical evidence that Sears was significantly less likely to hire female applicants, Sears rebutted the inference of discrimination by suggesting that female applicants themselves lacked interest in commission sales. The district court agreed and essentially found that “the company had merely honored the preexisting employment preferences of working women themselves.” Vicki Schultz and Stephen Petterson, Race, Gender, Work, and Choice: An Empirical Study of the Lack of Interest Defense in Title VII Cases Challenging Job Segregation, 59 U Chi L Rev 1073, 1077 (1992) (arguing that the validity of the lack of interest defense depends on the claim that women’s aversion to the position arose from social or cultural forces beyond the employer’s control).
  36. 42 USC § 2000e-2(k)(1)(B)(ii).
  37. Of course, no such changes would be needed if cases of intentional passive discrimination are rare or adequately captured by pattern and practice claims and one is unconcerned with both unintentionally induced workplace segregation and group-based differences in perceived or actual compensation and fringe benefits so long as no individual discrimination exists.

Post a Comment (all fields are required)

You must be logged in to post a comment.