On March 30, 2018, Judge Analisa Torres of the U.S. District Court for the Southern District of New York partially certified a class in Chen-Oster v. Goldman, Sachs & Co., a gender discrimination class action against Goldman, Sachs & Co. (“Goldman Sachs”). In so doing, Judge Torres not only departed from the Report and Recommendation of Magistrate Judge James C. Francis, but also extended beyond the U.S. Supreme Court’s reasoning in Wal-Mart Stores, Inc. v. Dukes, 564 U.S. 338 (2011).

Background of the Chen-Oster Litigation

On September 16, 2010, a group of female former Goldman Sachs employees (collectively, “Plaintiffs”) filed a putative class action alleging both disparate treatment and disparate impact gender discrimination claims pursuant to Title VII of the Civil Rights Act of 1964 and the New York City Human Rights Law. Plaintiffs sought to represent a putative class consisting of current and former female Associates and Vice Presidents employed by Goldman Sachs in revenue-generating divisions. The putative class covered between 1,762 and 2,300 employees nationwide. In general, Plaintiffs alleged that Goldman Sachs discriminated against class members in terms of evaluation, compensation, and promotion decisions.

Plaintiffs Targeted Numerous Discretion-Heavy Employment Practices

Plaintiffs targeted several practices that vested considerable discretion in supervisors and other evaluators. One such practice was Goldman Sachs’ 360 Reviews, which every employee underwent on an annual basis. After performing a self-evaluation, each employee designated eight to 12 evaluators (subordinates, supervisors, peers or internal clients), which could be modified by the employee’s direct supervisor. Then, the evaluators assigned a numerical value to the employee across a range of fixed criteria. At the end of the grading process, employees’ managers added their own evaluations and comments.

Plaintiffs also challenged Goldman Sachs’ annual “quartiling” process, in which managers were required, once a year, to assign their subordinates into one of five “quartiles” (25 percent of their direct reports in each of the top three quartiles; 15 percent in the fourth quartile; and 10 percent in the bottom quartile). Managers made the assignments based on a variety of considerations, including 360 Reviews, quality of performance, long-term commercial impact or contribution, technical and functional expertise, potential to assume increasing responsibility, leadership and management skills, and diversity and citizenship activities. Managers sometimes moved employees to different quartiles to enforce the distribution or avoid “grade inflation;” Goldman Sachs’ human resources department also inquired about and adjusted rankings if there were considerable discrepancies between employees’ 360 Reviews and their assigned quartiles.

For the majority of the putative class, their compensation consisted of a base salary and a bonus. After Goldman Sachs allocated a budget to each division, division-level managers further divided their respective budgets among their divisions’ business units. Managers would then recommend compensation levels for specific employees, which were reviewed at various levels back up the chain—by business unit heads, division-wide compensation committees, and finally by the firm-wide compensation committee.

Promotion decisions at Goldman Sachs were based on a process known as “cross-ruffing,” in which the human resources department worked with business unit heads and other managers to develop a list of promotion candidates from the top two quartiles. After being approved by the division heads, the lists were reviewed by a firm-wide talent assessment group. The human resources department then developed a list of “cross-ruffers:” Managing Directors at the firm who evaluated the candidates. Cross-ruffers were assigned candidates to review from outside of their business units and, in turn, were responsible for interviewing a number of employees familiar with the candidates’ work, using what Judge Torres referred to as “standardized forms and criteria.” Cross-ruffers met as a group and developed a ranked list of promotion candidates, which was then reviewed by the firm-wide talent assessment group and division heads. Promotion decisions were ultimately made by Goldman Sachs’ management committee.

The Court Found A “Common Mode” And “Significant Proof” Of Disparate Treatment, Despite Significant Managerial Discretion

As Judge Torres noted, in Dukes, the Supreme Court decided that a policy of allowing discretion by local supervisors over employment decisions was, on its face, the opposite of a uniform employment practice that would provide the commonality needed to justify certification. However, seizing upon language in the majority’s opinion, Judge Torres noted that a “common mode” of exercising discretion that pervaded an entire company could constitute a general policy, which, if supported by “significant proof” of disparate impact, could demonstrate the commonality necessary for certification.

Goldman Sachs argued that Plaintiffs failed to show that managers uniformly applied the 360 Review, quartiling, and cross-ruffing processes in making compensation and promotion decisions and that Plaintiffs could not do so because individual managers (in exercising their discretion) applied these processes in individualized ways. Judge Torres disagreed.

The court distinguished Dukes on the basis that the defendant in that case (Wal-Mart) had no common job evaluation procedure; employees had been subject to a variety of regional policies and practices. Thus, even though Goldman Sachs’ 360 Review, quartiling and cross-ruffing decisions involved a variety of individualized and highly specific evaluations, the court found they nonetheless constituted a “common mode” because there was no dispute that the processes themselves were employed across the class. Relying on one of Plaintiffs’ experts’ reports, which had aggregated data across business units, and over Goldman Sachs’ objections, the court next concluded that there was significant proof of disparate impact.

Employing A Similarly Lenient Analysis, The Court Also Found Common Issues On Plaintiffs’ Disparate Treatment Theories

The court also found Plaintiffs had established commonality with respect to their disparate treatment claims, which were based on both statistical and anecdotal evidence. Judge Torres found that the same statistical data used to support Plaintiffs’ disparate impact claims supported their disparate treatment theory. Next, the court found common evidence of gender bias, relying on anecdotal evidence of internal complaints, external complaints, survey answers, emails, articles, business records, and declarations from putative class members—including evidence of purported sexual assaults or harassment, verbal abuse, gender favoritism, stereotyping and retaliation.

The Court Held That Common Issues Predominated As To Plaintiffs’ Statistical Claims But Not Their Anecdotal Ones

After determining that the Plaintiffs were both typical and adequate, Judge Torres next evaluated whether Plaintiffs had satisfied the predominance requirement of Rule 23(b)(3). With respect to Plaintiffs’ disparate impact claim, Goldman Sachs could have defended against the claims on the merits by establishing that the challenged processes (i.e., 360 Reviews, quartiling and cross-ruffing) were job-related and consistent with business necessity. While the Magistrate agreed with the company that individualized inquiries regarding whether those processes caused individual class members cognizable injuries overwhelmed any predominance, Judge Torres disagreed. Without citing authority, Judge Torres took the position that the relevant question was whether Goldman Sachs could demonstrate the business necessities “of the processes,” in general, which the court deemed an issue capable of generalized proof.

As to Plaintiffs’ statistics-based disparate treatment theory, the court restated its position that Plaintiffs’ statistical evidence of a disparate impact raised an inference of discriminatory animus. The court, then, summarily stated that Goldman Sachs could rebut that inference by either challenging the accuracy or adequacy of Plaintiffs’ statistics or using non-statistical evidence to show a lack of discriminatory intent.

Finally, Judge Torres agreed with Goldman Sachs and the Magistrate that Plaintiffs’ anecdotal evidence-based disparate treatment theory did not satisfy Rule 23(b)(3)’s predominance requirement. Specifically, because this claim was based on a collection of varied specific instances of purported discriminatory behavior, Goldman Sachs’ defenses would likely go to the considerations and intentions of individual managers and/or business units, which would involve a predominance of individual issues.

Takeaways From Judge Torres’ Certification Order

Judge Torres’ order—which did not delve into the merits of Plaintiffs’ claims—may ultimately be challenged through an interlocutory appeal (and potentially reversed thereafter). However, Chen-Oster may nonetheless be instructive for other employers in evaluating their employment practices and pay and promotion statistics. To mitigate some of the risks of class certification, employers should examine their employment processes to assess the level of uniformity (or lack thereof) among hiring, evaluation and compensation procedures. Moreover, in the wake of increased legislation and litigation aimed at gender-based wage disparities, employers may wish to retain counsel to conduct internal pay audits to identify and address any pay disparities in their workforces.