Since 2013, federal government contractors have been subject to Directive 307 (DIR 2013-03), the Obama-era guidance document purporting to explain the approach of the Office of Federal Contract Compliance Programs (OFCCP) in analyzing compensation during compliance audits pursuant to Executive Order 11246. But Directive 307 was hardly a model of clarity and left contractors largely in the dark about the Agency’s statistical and legal methodology and standards for inferring compensation discrimination. Accordingly, contractors struggled to take proactive steps to anticipate findings by OFCCP and, if selected for an audit, often found themselves facing “indicators” of pay bias they did not fully understand and could not effectively rebut.
On Aug. 24, OFCCP officially rescinded Directive 307. In its place, OFCCP issued DIR 2018-05 (Analysis of Contractor Compensation Practices During a Compliance Evaluation). According to the Agency, this new directive “reinforces OFCCP’s commitment to greater transparency, consistency and efficiency in compliance evaluations.” But OFCCP’s pursuit of “consistency” and “efficiency” may well lead compliance officers to double down on a one-size-fits-all approach to compensation reviews, rather than embracing the nuanced, case- and company-specific analysis required by the governing law that DIR 2018-05 states OFCCP will (and must) follow. This article highlights key features of OFCCP’s approach, analyzes the ways in which it may conflict with Title VII standards, and considers the import of the directive for pending and future enforcement actions by the Agency.
What DIR 2018-05 tells us about OFCCP’s approach
DIR 2018-05 makes good on OFCCP’s promise to provide more “fulsome guidance” and “clarity” regarding how the Agency will approach evaluations of contractors’ compensation practices. The directive provides additional detail and transparency about the specific types of statistical techniques the Agency will use to round out its model and generate its results, for example. That transparency may be welcome news to contractors, as it appears to illuminate the approach to compliance audits that OFCCP has been utilizing for years. DIR 2018-05 likewise directs that compliance officers “must not begin the desk audit analysis of the contractor’s compensation data” until it has in hand all of the necessary information, which underscores the need to thoughtfully review data submissions before jumping into analysis. This additional direction may ultimately reduce “false positive” discrimination findings attributable to incomplete information or misunderstandings about what the contractor’s data does (and does not) mean. Further, by echoing the direction in a previous directive (DIR 2018-1) that OFCCP will issue Predetermination Notices rather than Notices of Violation, OFCCP signals that it will take contractor’s arguments into account before rushing to enforcement actions.
While DIR 2018-05 appears designed to ensure that both the contractor’s data and OFCCP’s methods are better understood, questions remain regarding whether the methodology the Agency intends to utilize will, in fact, lead to more accurate analysis of pay discrimination. DIR 2018-05 states at the outset that OFCCP “follow[s] Title VII,” and correctly acknowledges that under this statute “courts consistently hold that there is no single way to prove discrimination.” The attachment to the directive detailing the Agency’s approach, however, proclaims a singular reliance on “statistics to evaluate contractor pay practices under Executive Order 11246.” Although the directive states that the Agency may also seek other types of non-statistical evidence, it strongly suggests that the Agency will impose a statistical model in every compliance review it undertakes.
DIR 2018-05 states that OFCCP will make “comparisons of similarly-situated employees in its analysis,” and in turn defines similarly-situated employees as “those who would be expected to be paid the same based on: (a) job similarity (e.g., tasks performed, skills required, effort, responsibility, working conditions and complexity); and (b) other objective factors such as minimum qualifications or certifications.” A similar definition had appeared in Directive 307. Yet both the old and the new directive go on to authorize OFCCP to construct statistical models that may not adhere to this definition or make the relevant comparisons.
For example, DIR 2018-05 explains that OFCCP will construct “pay analysis groups” (or PAGs), another holdover term from the prior directive found nowhere in Title VII or its interpreting case law. These PAGs lump together employees according to “a contractor’s compensation system”—for example, the job families in place at a company—and the Agency states that it will include employees in the same PAG if they “perform[ ] broadly similar job functions.” From there, DIR 2018-05 states the Agency will introduce statistical controls to reflect “structural differences among members of the PAG” (such as division, product line, or location), as well as “individual employee characteristics related to the contractor’s pay determinations” (such as prior experience or education). The directive is silent, however, as to how OFCCP compliance officers should proceed if the investigation reveals situations where pay factors are not necessarily captured in a format easily amenable to statistical analysis—for example, where an employer differentiates pay based on the types of legitimate factors acknowledged by OFCCP, but does not systematically maintain them in a quantitative form (such as where performance reviews are narrative rather than numeric, or where different managers across separate divisions utilize different processes or practices to manage performance). Absent the ability to control for these additional factors, the Agency’s model may incorrectly make comparisons among employees grouped together in a PAG who only share “broadly similar job functions,” rather than the level of job similarity required to truly make individuals comparable under Title VII.
DIR 2018-05 in turn appears to suggest that whenever OFCCP’s statistical model generates bottom-line compensation disparities, contractors can expect “preliminary discrimination findings.” In other words, the new directive may indicate that OFCCP has transformed the law of compensation discrimination under Title VII into a statistical model. Although the Agency will provide the bases for its preliminary findings in the Predetermination Notice, DIR 2018-05 makes clear the explanation will take the form of “PAGs and regression results.” Greater transparency about this process may be cold comfort to a contractor who may find itself accused of systemic, intentional pay bias solely based on statistics.
The law on compensation discrimination requires more than simple statistics
Undisputedly, Title VII furnishes the legal standard for EO 11246. But the relevant Title VII cases dictate an approach that is significantly more nuanced than that set forth by the new directive. At the outset, DIR 2018-05 acknowledges that pay comparisons must be between “similarly situated” employees. Yet it goes on to bypass the careful, case-by-case evaluation of actual skills, duties, and responsibilities that Title VII demands. See, e.g., Warren v. Solo Cup Co., 516 F.3d 627, 630-31 (7th Cir. 2008) (rejecting Title VII compensation claim where plaintiff and her comparator were not “similarly situated” because they had “very different educational backgrounds, experiences, and qualifications”); Coser v. Moore, 739 F.2d 746, 753 (2d Cir. 1984) (affirming rejection of Title VII compensation discrimination claim because plaintiffs’ statistical analysis did not account for differences in prior work experience and actual job duties, which were “crucial variables that must be taken into account in comparing relative salaries”). Indeed, in Title VII compensation discrimination litigation, courts routinely dismiss statistical analyses that fail to compare employees who are similarly situated. See, e.g., Anderson v. Westinghouse Savannah River Co., 406 F.3d 248, 263 (4th Cir. 2005) (affirming exclusion of statistical analysis that lumped together all employees into overbroad job groupings, finding that “there [was] simply too much disparity in the groups” and no “factor that would control for the actual job title or the job duties”); Cooper v. S. Co., 260 F. Supp. 2d 1305, 1314, 1317 (N.D. Ga. 2003) (striking expert report where the statistical analysis “fail[ed] to compare similarly situated individuals” because it did not “account for differences in the type or level of the employees’ applied skills, both of which are highly related to compensation” and did not “compar[e] employees with equivalent work experience in specific job categories or job progressions”), aff’d, 390 F.3d 695 (11th Cir. 2004).
Additionally, DIR 2018-05 does not fully embrace Title VII’s requirement that the government prove either a pattern or practice of intentional discrimination or a disparate impact generated by a uniform neutral employment practice. See, e.g., United States v. City of N.Y., 717 F.3d 72, 87 (2d Cir. 2013) (underscoring the “high bar for the prima facie case the Government or a class must present in a pattern-or-practice case,” which requires “evidence supporting a rebuttable presumption that [the employer] acted with the deliberate purpose and intent of discrimination against an entire class”); Watson v. Fort Worth Bank & Trust, 487 U.S. 977, 994 (1988) (to prove a disparate impact claim, a plaintiff must “go[ ] beyond … show[ing] that there are statistical disparities in the employer’s work force” and identify the “specific employment practice” alleged to have caused a disparate impact). Instead, the directive appears to treat any compensation differentials generated by the Agency’s statistical model as per se proof of discrimination. But case law routinely acknowledges that bare statistical evidence rarely—if ever—suffices on its own. See, e.g., Gay v. Waiters’ & Dairy Lunchmen’s Union, 694 F.2d 531, 552-53 (9th Cir. 1982) (““[The question whether the facts proved are sufficient to permit a legal inference of discriminatory intent cannot properly be reduced into a mere battle of statistics.… Simply put, statistics demonstrating that chance is not the more likely explanation are not by themselves sufficient to demonstrate that race is the more likely explanation for an employer’s conduct.”). Accordingly, contractors facing a discrimination finding based on statistical disparities may need to look behind the Agency’s pronouncements to the governing law.
The attachment to DIR 2018-05 purports to recite the law of disparate treatment and disparate impact under Title VII. It cites only two cases, however—International Brotherhood of Teamsters v. United States, 431 U.S. 324 (1977) and Griggs v. Duke Power Co., 401 U.S. 424 (1971)—and neither of these involved compensation claims. These forty-plus year-old decisions each concerned alleged hiring discrimination for relatively low-skill jobs (at a trucking company and power plant, respectively). With respect to hiring for such positions, one might expect to see members of different racial groups hired in rough proportion to their representation in the pool of qualified applicants, and to be able to determine who is “qualified” by reference to a small number of objective characteristics. Indeed, in Teamsters and Griggs, the Supreme Court found that evidence showing that the company hired a disproportionately low number of African-Americans—none, in fact, in Teamsters—for the positions at issue was probative of discrimination. Similarly in Griggs, the court found probative evidence that a specific employment practice (the requirement of a high school degree, which was not actually needed for the job at issue) excluded a disproportionate number of African-Americans from consideration.
But the use of simple statistics to assess those types of straightforward hiring claims does not support similar reliance on a single statistical model to capture and control for the entire range of factors that legitimately influence compensation decisions. The flaws with this analogy are particularly stark in industries with highly skilled and/or highly differentiated workers, or where a single statistical model groups together a broad range of different positions for which different skill and performance factors matter. In these types of cases, one would not expect a simple statistical model to capture the richness of the factors that differentiate employees over time, the value of different departments or teams to a company’s performance, or the myriad and fluid market factors that can lead different workers to command different salaries.
Moreover, the Bennett Amendment to Title VII incorporates the four Equal Pay Act (“EPA”) affirmative defenses, including the defense of “any other factor than sex [or race].” Consequently, in a compensation discrimination case, if a practice satisfies one of the affirmative defenses of the EPA, it cannot constitute a violation of Title VII. See, e.g., Maxwell v. City of Tucson, 803 F.2d 444, 446 (9th Cir. 1986). The necessity to consider these affirmative defenses—including primarily the fourth “catch-all” defense—in any assessment of compensation discrimination further differentiates compensation claims from hiring claims, and illustrates why case law involving pattern-or-practice hiring claims is distinguishable from, and less complicated than, cases addressing compensation.
DIR 2018-05 does cite to the EEOC’s 2000 Compliance Manual Chapter 10: Compensation Discrimination for the proposition that compensation analyses must compare employees who are “similarly situated,” such that one would “expect them to be paid the same.” That chapter confirms that determining who is similarly situated is a fact-intensive inquiry that requires looking at “the actual content of jobs” to determine whether they “involve similar tasks, require similar skill, effort, and responsibility, and are similarly difficult or complex.” The manual—in accordance with Title VII law—explicitly acknowledges that job titles and descriptions are “not controlling,” and confirms one cannot assume two employees with the same job title are similar under the law; some employees with different job titles may in fact perform substantially similar work (e.g., maids versus janitors), whereas employees with a shared title may not (e.g., professors across a university with different research and teaching specialties and focus). And the manual underscores that even where a pay differential exists between employees who are truly similarly situated in terms of the work they perform, one must then ask whether there is “a non-discriminatory reason” that explains the difference. Performance, relevant prior experience, and seniority are a few of the examples the manual provides. Because the inquiry is varied and variable, the manual concludes that “[t]he decision about whether and how to use statistics” to evaluate whether pay discrimination exists “should be made on a case by case basis.”
While DIR 2018-05 fulfills OFCCP’s stated goal of providing more transparency to contractors, it also illuminates a departure from the case-by-case approach articulated in the EEOC manual, and Title VII law more generally. Where Title VII requires a fact-intensive analysis of particular jobs, DIR 2018-05 advises OFCCP to create aggregated, dissimilar “pay analysis groups.” Where Title VII mandates that an employer be afforded the opportunity to explain any pay differentials based on factors and facts particular to the positions and employees at issue, DIR 2018-05 suggests that only company-wide factors that OFCCP can layer onto its aggregated statistical model suffice. And where even the EEOC manual counsels a “case by case” consideration of whether statistical evidence is even probative, DIR 2018-05 endorses a one-size-fits-all approach.
Will Administrative Law Judges and courts defer to DIR 2018-05 nonetheless?
If OFCCP believes a contractor has engaged in compensation discrimination and cannot resolve those allegations to its satisfaction through the mandatory conciliation process, it can bring an enforcement action before an Administrative Law Judge (ALJ) with the U.S. Department of Labor. The ALJ will adjudicate the OFCCP’s claims in the first instance. If the contractor is dissatisfied with the ALJ’s recommended decision resolving OFCCP’s claims of discrimination, the contractor can seek review by the DOL’s Administrative Review Board and, ultimately, review in federal court (indeed, de novo review on issues of law). Accordingly, the extent to which decision-makers can or must defer to DIR 2018-05, particularly if it departs from Title VII standards in important ways, matters.
DIR 2018-05 was issued without any formal notice-and-comment rulemaking process, and states that it “does not change the laws and regulations governing OFCCP’s programs and does not establish any legally enforceable rights or obligations.” Because the directive is merely an agency “enforcement guideline,” it should be reviewed by courts without the high level of deference under Chevron v. Natural Resources Defense Council, Inc., 467 U.S. 837 (1984), and followed (if at all) only to the extent that it is correct—i.e., to the extent that it “ha[s] the ‘power to persuade.’” Christensen v. Harris Cty., 529 U.S. 576, 587 (2000) (quoting Skidmore v. Swift & Co., 323 U.S. 134, 140 (1944)); accord Vance v. Ball State Univ., 570 U.S. 421, 431 n.4 (2013). Indeed, DIR 2018-05 states that it is merely “intended to provide guidance to OFCCP staff or federal contractors on enforcement and compliance policy or procedures” and that its “purpose” is “[t]o outline standard procedures for reviewing contractor compensation practices during a compliance evaluation”—not to establish agency policy.
Going forward, contractors can expect OFCCP compliance officers to use the formulaic approach that DIR 2018-05 outlines, and may find themselves subject to evaluations that would reduce the complicated law of Title VII compensation discrimination to a numbers game. But contractors should not accept that this approach in fact comports with the Agency’s mandate or contractors’ obligations under EO 11246 to refrain from discrimination on the basis of sex, race, or other protected characteristics. Instead, contractors may need to arm themselves with a rich understanding of what Title VII requires, and insist that the government faithfully execute its enforcement authority consistent with the law.
Erin Connell is a partner in the San Francisco office of Orrick, Herrington & Sutcliffe, specializing in major employment litigation, including issues related to pay equity, OFCCP investigations, EEOC matters and systemic discrimination guidance.
Gary Siniscalco is senior counsel in Orrick’s San Francisco office and has co-led the firm’s EEO and OFCCP practice groups. An experienced adviser on EEOC and Department of Labor matters, he served as regional counsel and senior trial attorney for the EEOC in San Francisco before joining Orrick.
Chris Wilkinson, a partner in Orrick’s Washington, D.C. office, specializes in labor and employment matters, previously serving as Associate Solicitor for Civil Rights and Labor Management for the U.S. Department of Labor.
Kathryn Mantoan is a senior associate in Orrick’s Portland office, and is a member of the firm’s employment practice specializing in class actions and issues involving pay equity.