Kassman v. KPMG LLP, No. 1:2011cv03743 - Document 831 (S.D.N.Y. 2018)

Court Description: OPINION AND ORDER: For the foregoing reasons, Plaintiffs motions for Rule 23 class certification and EPA collective certification are DENIED. Both parties Daubert motions and motions to strike are DENIED. The Clerk of Court is directed to close the motions at Dkt. Nos. 797 , 800 , 802 , 803 , 813 , 815 , 817 , 818 , 827 and 829 .(Signed by Judge Lorna G. Schofield on 11/30/2018) (jcs)

Download PDF

Kassman v. KPMG LLP UNITED STATES DISTRICT COURT SOUTHERN DISTRICT OF NEW YORK -------------------------------------------------------------X : DONNA KASSMAN, et al., : : Plaintiffs, : : -against: : KPMG LLP, : : Defendant. : -------------------------------------------------------------X Doc. 831 11/30/18 11 Civ. 3743 (LGS) OPINION AND ORDER LORNA G. SCHOFIELD, District Judge: Plaintiffs bring this putative sex discrimination class action against Defendant KPMG LLP (“KPMG”). Plaintiffs assert claims under the disparate impact and disparate treatment provisions of Title VII of the Civil Rights Act of 1964 on behalf of a nationwide class of more than 10,000 female KPMG employees, from 2009 to the present, and analogous claims under New York state and city law on behalf of a New York subclass. Plaintiffs also seek second stage collective action certification of the Equal Pay Act claims of 1,112 Opt-In Plaintiffs. Plaintiffs filed this suit on June 2, 2011, claiming that KPMG discriminates against thousands of women in their pay and promotions. Less than three weeks later, the Supreme Court decided Wal-Mart Stores, Inc. v. Dukes, 564 U.S. 338 (2011) (“Dukes”). Dukes makes it extremely difficult for a gender discrimination suit to proceed as a class action when the discriminatory treatment was the product of local supervisors exercising their discretion in awarding pay and promotions. See id. at 355. Most of the conduct challenged in this lawsuit occurred after Dukes provided a roadmap to avoid class certification of a nationwide class asserting gender discrimination. Not surprisingly, during the years since Dukes, with this lawsuit pending, KPMG has utilized a decentralized system for determining pay and promotion reminiscent of that used by Walmart in Dockets.Justia.com Dukes. KPMG also has been attentive to gender disparities and workplace misconduct. (It would have been surprising -- indeed irresponsible -- if KPMG had not been attentive to gender issues with potential liability looming these past years.) Although individual decisions are made by managers at the local level, KPMG’s overall system is not the product of accident or happenstance. It is under the direction of a National Director of Compensation Strategies, who -- with a staff of approximately 12 people -- is responsible for designing and implementing the firm’s employee compensation program and managing the firm’s performance recognition programs. The result, according to Plaintiffs, is a pay disparity between women and men of approximately 2.8%. Although any statistically significant pay disparity on account of gender would be improper, for context, the national disparity in pay between women and men is 18%. See Nikki Graf, Anna Brown and Eileen Patten, The Narrowing, but Persistent, Gender Gap in Pay, Pew Research Center (“In 2017, women earned 82% of what men earned . . . .”) (last visited November 27, 2018).1 KPMG counters Plaintiffs’ analysis and asserts that Plaintiffs have provided “no statistical evidence of anything more than sporadic and isolated within-job sex disparities in pay.” Regardless of who is correct, under Dukes, the proposed class cannot be certified; Plaintiffs’ argument, at its core, still boils down to “managers, who were left without meaningful guidance . . . , fell back on their own stereotyped views of women in making pay and promotion decisions.” Dukes v. Wal-Mart Stores, Inc., 964 F. Supp. 2d 1115, 1127 (N.D. Cal. 2013) (“Dukes II”) (on remand). The district court on remand in Dukes called this “a perfectly logical theory” for liability, id., an observation echoed by at least one other court, see Moussouris v. Available at: http://www.pewresearch.org/fact-tank/2018/04/09/gender-pay-gap-facts/. 2 Microsoft Corp., No. 15 Civ. 1483, 2018 WL 3328418, at *15 (W.D. Wash. June 25, 2018). Nevertheless, this argument “leaves [p]laintiffs right back where they started: challenging [defendant’s] practice of delegating discretion to local managers, which the Supreme Court specifically held was not a specific employment practice supplying a common question sufficient to certify a class.” Dukes II, 964 F. Supp. 2d at 1127. As Plaintiffs provide insufficient evidence of “some glue” holding together the reasons for the countless individual employment decisions they challenge, the motion for class certification is denied.2 Dukes, 564 U.S. at 352. I. BACKGROUND The facts below are taken from the parties’ submissions in connection with the pending motions, and the Court resolves factual disputes as necessary for the disposition of those motions. See Shahriar v. Smith & Wollensky Rest. Grp., Inc., 659 F.3d 234, 251 (2d Cir. 2011). A. Plaintiffs and the Proposed Class Plaintiffs are Donna Kassman, Sparkle Patterson, Jeanette Potter, Ashwini Vasudeva, Tina Butler, Cheryl Charity, Heather Inman, Nancy Jones and Carol Murray. The proposed nationwide class consists of female Associates, Senior Associates, Managers, Senior Managers/Directors and Managing Directors employed within KPMG’s Tax and Advisory 2 One wonders, however, about the empirical reality of a key assumption that underlies Dukes -that “left to their own devices most managers in any corporation -- and surely most managers in a corporation that forbids sex discrimination -- would select sex-neutral, performance-based criteria for hiring and promotion that produce no actionable disparity at all.” Dukes, 564 U.S. at 355. This assumption is contradicted by studies finding that implicit bias has profound effects. See, e.g., Dukes, 564 U.S. 338 at 373 n.6 (2011) (Ginsburg, J., dissenting) (citing Claudia Goldin & Cecilia Rouse, Orchestrating Impartiality: The Impact of “Blind” Auditions on Female Musicians, 90 Am. Econ. Rev. 715, 715–16 (2000)); see generally Annika L. Jones, Implicit Bias as Social-Framework Evidence in Employment Discrimination, 165 U. Pa. L. Rev. 1221, 1228 (2017) (Before Dukes, “[c]ourts generally accepted . . . even generalized evidence of implicit bias to supply the ‘glue holding the class theory together.’”). 3 Functions between October 30, 2009, through the date of judgment (the “Class Positions” and “Class Period”). 3 Plaintiffs held Class Positions during the Class Period. B. Management Structure of KPMG KPMG is a professional services firm with more than 34,000 employees in over 90 offices nationwide. KPMG is governed by a Board of Directors, composed of up to 18 of KPMG’s top partners, which is responsible for adopting firm-wide policies and overseeing the firm’s management. The firm’s Management Committee implements firm policies as promulgated by the Board. KPMG is divided into four Functions, three with client-facing professionals -- Advisory, Tax and Audit -- and the Business Process Group, which provides back-office support. The Management Committee includes leaders from each Function. Each Function is led by a National Managing Partner, who reports to a Vice Chair, who reports to the Management Committee. This case involves only the Advisory and Tax Functions, which together employ more than 10,000 women in Class Positions. These Functions together have over 150 specialized practice areas that provide a variety of services to businesses across a broad range of industries and locations. Advisory is divided into 4 Service Groups -- Management Consulting, Risk Consulting, Deal Advisory and Strategy -- which are subdivided into approximately 15 Service Lines, and then increasingly more focused practice areas -- over 40 Service Networks, over 80 sub-practice areas (called “Cost Centers”) and numerous Service Offerings, which are targeted toward particular industries. Tax has 6 Service Lines, subdivided into nearly 70 Cost Centers and 3 Excluded from the class are employees in KPMG’s client services support groups, such as Human Resources; Information Technology; Facilities; Finance; and Sales, Marketing and Communications. 4 other sub-practices. Each sub-practice has a Partner leader (or Principal if they do not have an active CPA license) and budget, down to at least the Service Network in Advisory or Cost Center in Tax. KPMG’s varied service offerings and industries result in diverse skills, responsibilities and working conditions across jobs. The firm’s professionals include, for example, accountants, scientists, engineers, lawyers and Ph.D. economists. The hierarchy of non-Partner positions at KPMG beginning with the most junior is: Associates; Senior Associates; Managers; Senior Managers/Directors and Managing Directors. C. Compensation During the Class Period, KPMG’s overall compensation strategy was to compensate at the market rate, differentiating among employees based on performance. KPMG calibrates the 75th and 25th percentiles of its compensation for each position with the corresponding percentiles in the market. Employee compensation consists of two components, annual salary increases (called “merit increases”) and variable compensation (bonuses). Both are determined through KPMG’s annual compensation process. Compensation budgets are determined on a firm-wide basis each year. The process begins with the Vice Chairs and National Managing Partners for each Function creating financial projections and proposing a compensation budget, each of which is sent to KPMG’s firm-wide Compensation Strategies Department. The Compensation Strategies Department then conducts a market study to determine the market compensation for the various jobs. The Compensation Strategies Department uses the market study and budgets to create proposed salary ranges for each job within each practice group. The Compensation Strategies Department then conducts salary range meetings with the senior management of Tax and 5 Advisory, which can result in revisions to the proposed ranges. The Management Committee approves the compensation budgets before KPMG begins the process of determining individual employees’ merit increases and variable compensation. After the salary ranges and budgets are approved, the Compensation Strategies Department provides benchmark merit increase and variable compensation figures for all of the positions throughout the firm via a firm-wide “Compensation Tool.” Merit increases and variable compensation are expressed in the Compensation Tool as percentages of current base salary, as are pay raises for promotions. The tool recommends percentage ranges for merit increases and variable compensation, with percentage bands based on the employee’s performance level (as determined at the annual assessment meeting) and current salary in relation to the market. The percentage bands overlap such that an employee with a relatively lower performance rating who is undercompensated compared to the market can get a higher percentage increase than someone with a higher rating who is compensated more favorably compared to the market. The Compensation Tool also generates a “target” salary. The “target” is the firm’s estimate of market-rate compensation for a particular employee based on factors such as job content, geography, performance and tenure. KPMG endeavors to bring all employees’ compensation toward their target compensation over time, amortizing any premiums paid to lateral hires over several years and keeping homegrown talent up to pace with the market. As stated in a Tax compensation presentation, “Over time, we expect employees to move closer to their relative reference point and we recognize that this may not occur in a single Salary Review process.” Sub-practice area leaders -- who are usually Partners, but can be a Managing Director, if they are the most senior professional in a given practice area and office -- ultimately determine compensation and receive the prepopulated salary increase and variable compensation 6 benchmarks for individual employees. The sub-practice area leaders review the benchmarks and use their discretion to make any appropriate adjustments for individual employees. These subpractice area leaders participate in compensation training programs that instruct them on the firm’s compensation policies, and warn them to avoid gender biases in determining pay. KPMG tells decision makers that they are not bound by the prepopulated benchmarks and expects that they will deviate from the benchmarks based on their judgment. As explained in an Advisory compensation training presentation, “[W]hen we develop the guidelines, we don’t have line of sight into the individual and unique contributions of each employee so we count on you to apply those guidelines in a way that makes most sense for your employees . . . .” Similarly, a Tax presentation states: “[Benchmark] values do not represent minimums, maximums, or exact prescribed salaries -- they are simply a general point of reference. . . . It should be noted that in some geographies and practices not all employees’ compensation is within the anticipated relative range of the compensation tool values.” While the sub-practice leaders have discretion to determine individual salary increases and variable compensation, allocating these amounts among employees in a practice area is a zero-sum exercise; the sub-practice leaders’ total allocation of compensation within their practice area needs to align with the budget. Sub-practice leaders’ proposed compensation allocations are subject to review against budget. Ultimately, the Vice Chair, the National Managing Partner and the National Head of Human Resources for each Function execute final approval of the compensation budget. D. Performance Assessment and Promotions Assessment of employees’ performance takes place mainly at the practice area level. Each professional employee in Tax and Advisory has a People Management Leader (“PML”), who is a more senior employee (i.e., Manager or higher) typically in the same practice. The PML 7 works with each “counselee” to set individual performance goals. Each employee’s performance against goals is assessed both at a mid-year review meeting and at a year-end assessment meeting held by the employee’s individual practice area, resulting in hundreds of such meetings each year across the many Tax and Advisory practices. The employee’s PML, other senior employees with knowledge of their work and a designated human resources (“HR”) representative participate. The scope of the assessment meeting varies by practice area and may cover, for example, one practice in one office, a sub-practice in a few offices, an entire practice area nationwide or a regional subgroup of a Service Network. Local practice leaders determine the structure, attendance, procedure and assessment criteria. For example, Tampa BTS at one time ranked all their employees -- from Associates to Senior Managers, regardless of job level -- in a single list in order of “which one could you not live without in this practice?” On the other hand, San Francisco Accounting Advisory Services ranks its employees only within job levels. The Pacific Southwest region of the IAER Service Network does not rank, but discusses each employee and notes low performers. KPMG’s performance rating metric has changed over the relevant time in this action -from 2009 to the present. Prior to 2011, a “9-box” system was used. From 2011 through 2015, a 5-point rating system was used. In 2016, KPMG began using the relative percentile of an employee’s variable compensation (bonus) award in the applicable range to denote performance. In other words, higher-rated performers received a higher percentage of their base salary as a bonus than lower-rated performers. Practice groups also keep track of performance distinctions among employees who receive the same rating or bonus percentile. Partners at more senior levels 8 and firm leadership may review aggregate performance and pay data, but generally do not review individual performance, compensation or promotion decisions.4 To be eligible for promotion at KPMG, an employee must have worked a certain number of years at a particular position, or have commensurate external experience. Tenure or experience requirements may be reduced by high performance ratings or advanced degrees. For instance, as of 2013, to be promoted from Manager to Director in Advisory, an employee should generally have at least eight years of experience in professional services, three to five years of management experience and have spent at least three years as Manager unless rated 1 or 2 (out of 5) for two consecutive years. A master’s degree may qualify as one year of experience, and a J.D. may qualify as two years. Employees received documents describing qualitative criteria for promotion readiness. For instance, the criteria for promotion to Senior Manager in Tax include reputation, thought leadership, project management and execution, risk management, recruiting, team management, communication skills, coaching and mentoring, understanding client issues, relationship building, personal brand, external involvement and profitable growth opportunity. PMLs and other practice area members discuss an employee’s promotion readiness with the employee as a matter of course at several points throughout the performance management process, including (1) at the annual goal-setting meeting with the PML, (2) prior to mid-year review meetings and (3) prior to year-end assessment meetings. Promotion readiness is also evaluated for each eligible employee at the year-end practice area assessment meetings. Promotions are ultimately determined by consensus at the annual practice group meetings, which 4 Although Plaintiffs dispute this, they have not presented credible evidence that shows otherwise. 9 are conducted by individual practice area members. The parties proffer no evidence that senior management reviewed individual promotion decisions to positions lower than Managing Director. E. Job Responsibilities Declarations submitted by 24 KPMG executives illustrate the wide-ranging educations, experiences, skills and specializations of employees across the 150 Cost Centers in Tax and Advisory. Examples from three Cost Centers within Tax are illustrative. Employees in Tax Controversy Services have J.D.s or L.L.M.s and represent clients in resolving disputes with the IRS. Employees in Economic Services have a Masters or Ph.D. in economics and create econometric and statistical models for KPMG clients. Employees in the Accounting Methods and Credits specialize in one type of tax credit or industry for which they have scientific or technical expertise, such as oil and gas, software or architecture. To study the disparate job responsibilities across KPMG, Defendant’s Industrial Psychologist, Dr. Cristina Banks, performed an “individualized job analysis” of “the job characteristics and work performed by KPMG employees in the Tax and Advisory functions . . . .”5 Dr. Banks conducted structured interviews with 81 female employees across 5 The expert reports, and other documents discussed in this Opinion, are subject to a Protective Order agreed to by the parties. These materials are “judicial documents,” as they are “relevant to the performance of the judicial function and useful in the judicial process.” Lugosch v. Pyramid Co. of Onondaga, 435 F.3d 110, 119 (2d Cir. 2006); see also Bernstein v. Bernstein Litowitz Berger & Grossmann LLP, 814 F.3d 132, 139 (2d Cir. 2016) (“In determining whether a document is a judicial record, we evaluate the ‘relevance of the document’s specific contents to the nature of the proceeding’ . . . .”). To the extent that information in those documents is disclosed in this Opinion, the privacy and business interests that justified their confidential treatment during discovery are outweighed by “the public’s right of access to judicial documents necessary to understand the basis for court rulings.” Spinelli v. Nat’l Football League, 903 F.3d 185, 193 n.2 (2d Cir. 2018); see also Dodona I, LLC v. Goldman, Sachs & Co., 119 F. Supp. 3d 152, 155 (S.D.N.Y. 2015) (“[E]ven if material is properly designated as Confidential or Highly Confidential by a protective order governing discovery, that same material might not overcome the presumption of public access once it becomes a judicial document.”). 10 Tax and Advisory. The interviewees were randomly selected from a study sample, which was designed to mirror the population of Opt-In Plaintiffs based upon: (1) Cost Center; (2) job level; (3) office; (4) job title and (5) tenure in position. Dr. Banks informed each interviewee that the information collected would not be shared with supervisors or used for performance evaluation, and that participation would be neither rewarded nor penalized. Dr. Banks and her staff conducted all interviews using the same verbatim interview protocol. The interviewees were asked 94 open-ended questions, such as, “Do you typically work with large, medium or small-sized companies? In what industry or industries?” Dr. Banks then performed a qualitative content analysis, coding interviewee answers when possible.6 Dr. Banks computed frequency distributions for the 68 questions that could be coded. For example, with respect to business development, 14% of interviewees had formal responsibility for selling KPMG services, 46% had informal responsibility and 41% had no responsibility. Certain questions, however, could not be coded, because the responses were too varied to be sorted into coherent categories. For example, when asked about the specific knowledge required for doing their jobs, interviewees responded with more than 32 different answers. Based on her study, Dr. Banks concludes, “KPMG professionals differ widely in the jobs they perform in the Tax and Advisory functions of the firm . . . .” To demonstrate this variety, Dr. Banks’s report holistically describes a number of interviewees. For example, Dr. Banks’s report juxtaposes a Senior Associate in the Procurement and Product Operations Cost 6 Randomly assigned teams of two coders independently analyzed the same set of responses. If the coding did not match, the two coders would attempt to reach a consensus, but if they could not agree, then a third coder would determine the final code. 11 Center of the Advisory function, who has an M.B.A. and Master’s in mechanical engineering and works directly with clients on implementation projects in the power and utility industries, with a Senior Associate in the Federal IA & SOAS Strategic Sourcing Cost Center of the Advisory function, who has a Bachelor’s in accounting and works with federal agencies to identify and mitigate financial risks. F. Sex-Based Disparities Both parties hired statistical experts to study sex-based disparities in pay and promotion at KPMG; they reach differing -- but not necessarily contradictory -- conclusions. As described below, Dr. Vekker and Dr. Bloom may both be correct; women in Class Positions were paid less than men who held the same job titles (per Dr. Vekker), but that discrepancy is associated predominantly with Service Line or Cost Center (per Dr. Bloom) -- i.e., the highest paying Cost Centers had more men and/or the lowest paying Cost Centers had more women. 1. Plaintiffs’ Expert Dr. Vekker Plaintiffs’ statistical expert, Dr. Alexander Vekker, analyzed whether women in the Class Positions were paid and promoted less than men who held the same positions at KPMG. His answer is yes. a. Pay Dr. Vekker ran aggregate linear regressions using KPMG data from 2008 to 2016 and found statistically significant differences in compensation -- both in terms of base pay and total pay -- between men and women who held Class Positions, when controlling for job level, experience, education, job location and performance ratings. “[M]easuring the significance of statistical disparities involves calculation of the standard deviation as a measure of predicted fluctuations from the expected value of a sample.” Burgis v. N.Y.C. Dep’t of Sanitation, 798 F.3d 12 63, 70 n.6 (2d Cir. 2015) (quoting Hazelwood Sch. Dist. v. United States, 433 U.S. 299, 308 n.14 (1977)). As the number of standard deviations increases, the probability that differences in pay could have been generated by chance decreases. Generally, results are deemed significant at around 2 standard deviations (i.e., 1.96), which signifies a probability of 5 in 100 (0.05) that the differences are due to chance. See Burgis, 798 F.3d at 70 n.6; see also Jones v. City of Boston, 752 F.3d 38, 46–47 (1st Cir. 2014) (describing 1.96 standard deviations as the threshold for significance most used by social scientists and adopted by “most federal courts” to measure disparate impact). Dr. Vekker’s regressions show that between 2008 and 2016, women in Class Positions in Tax earned an average of 2.7% less in total compensation, with a standard deviation of 7.88. Likewise, women in Class Positions in Advisory earned an average of 2.8% less in total pay, with a standard deviation of 8.19. b. Promotions With respect to promotions, Dr. Vekker ran logistic regressions, because the dependent variable is binary (i.e., promoted or not promoted). Dr. Vekker used the same controls as those for the compensation regressions but included an additional control for having a CPA license. He found that women in all Tax Class Positions were on average 3.4% less likely than men to be promoted, with a standard deviation of 7.64. When broken down by position, female Senior Associates in Tax were most disadvantaged, as they were 5.5% less likely than male Senior Associates to be promoted to Manager, with a standard deviation of 5.97. Women in Advisory Class Positions were an average of 1.8% less likely to be promoted than men, with a standard deviation of 4.93. Female Managers in Advisory were most disadvantaged, as they were 2.9% less likely to be promoted than men, with a standard deviation of 3.66. 13 2. Defendant’s Expert Dr. Bloom Defendant’s statistical expert, Dr. David Bloom, analyzed “whether there is meaningful statistically significant evidence of sex-based differences in pay and promotion -- adverse to women -- involving US-based [employees] holding job titles of Associate, Senior Associate, Manager, Senior Manager/Director, or Managing Director in KPMG’s Tax function or its Advisory function during 2009 through 2016.” Dr. Bloom concluded that no statically significant differences exist. a. Pay Dr. Bloom argued that Dr. Vekker’s aggregate regressions “erroneously equate[] job with job title,” leading them to “incorrectly pool[] data across job titles.” To support the proposition that Dr. Vekker erred by amalgamating all persons within one Class Position into a single category, Dr. Bloom pointed to divergent average base pay rates across Service Lines. For example, within Tax, an Associate in the Mergers and Acquisitions Service Line earned on average approximately 20% and 60% more than both Senior Associates and Associates respectively in Business Tax Services. This discrepancy between persons holding the same job title -- or even differing job titles -- was greater at the Service Network and Cost Center levels. For example, in Tax, a Managing Director in the SALT Income and Incentives Cost Center earned on average more than three times as much as a Managing Director in the BTS High Volume Trust Tax Cost Center. Similar but smaller differences existed in Advisory. Large discrepancies between persons with the same title sometimes existed even within the same Service Network, for example in the Restructuring Service Network. According to Dr. Bloom, Dr. Vekker’s failure to acknowledge these differences and instead consolidating, for example, all Senior Associates in Tax into a single category, regardless 14 of their specialized job and accompanying average pay, led Dr. Vekker to find a sex-based disparity in pay where none exists.7 Rather, Dr. Bloom’s regressions suggest that the gender pay disparity Dr. Vekker found reflects a heavier concentration of men in higher compensated units and heavier concentration of women in lesser compensated units. To illustrate that area of specialization, rather than sex, drives any pay disparity, Dr. Bloom repeated Dr. Vekker’s analyses, but, instead of running the regression by job title, he used a job title-Service Line combination. When accounting for Service Line in addition to job title, Dr. Bloom’s regressions yielded few statistically significant pay disparities. In Tax, the estimated sex disparities are either positive (i.e., favored women) or statistically insignificant for 96% of those year-title-service line combinations with a sample size of 30 or more, which cover 91% of the overall sample. Likewise, in Advisory, the estimated sex disparities are either positive (i.e., favored women) or statistically insignificant for 91% of those with a sample size of 30 or more, which cover 85% of the overall sample. Based on these regressions, Dr. Bloom concludes that there is “no statistical evidence of anything more than sporadic and isolated within-job sex disparities in pay” at KPMG. b. Promotions Dr. Bloom also repeated Dr. Vekker’s logistic regression, but without pooling data across job titles, using two different statistical methods.8 With respect to Tax, of the 224 (of 247 possible) reported analyses of promotion rates, the estimated sex disparities are either positive 7 Dr. Bloom also ran models to correct other perceived deficiencies in Dr. Vekker’s analysis, such as pooling data from 2008 to 2016 and failing to include years of experience at KPMG as a control variable. 8 Fisher’s Exact Tests are used to analyze small sample size, whereas logit results require sample sizes of at least 30. 15 (i.e., favored women) or statistically insignificant in 221 (99%). Additional logit estimates are reported for 135 of the 247 cells; of those 135 cells, 129 (96%) are either positive (i.e., favored women) or statistically insignificant. With respect to Advisory, of the 330 possible analyses of promotion rates, Fisher’s Exact Test results are reported for 287 cells. Of those 287 cells, the estimated sex disparities are either positive (i.e., favored women) or statistically insignificant in 282 (98%). Additional logit estimates are reported for 212 of the 330 cells. Of those 212 cells, 209 (99%) are either positive (i.e., favored women) or statistically insignificant. Based on these analyses, Dr. Bloom concluded that there was no firm-wide promotion disparity at KPMG.9 G. Work Environment The parties provided evidence supporting contrasting narratives of the work environment for women at KPMG. Margaret Teegan, the Managing Director of Human Resources Communications and Culture, submitted a declaration describing KPMG’s “long history of inclusion and diversity efforts,” noting awards that the firm has received, such as the “prestigious Catalyst Award” for “proven, measurable results that benefit women across a wide range of dimensions.” All employees are governed by KPMG’s Code of Conduct, which espouses “KPMG’s core values” such as “respect the individual,” and requires them to “‘raise their hand’ and report any observed or potential violation.” The Code of Conduct does not include a policy 9 Dr. Vekker responded to Dr. Bloom’s criticisms of his analyses by arguing that Dr. Bloom exhibits an “indifference to statistical power” by disaggregating the pooled data, which is necessary to have a sufficiently large sample size. See Moussouris, 311 F. Supp. 3d at 1236 (“[D]isaggregating data as Microsoft suggests seems to produce pools with varying sizes, some of which may ‘mask common mechanisms’ because of their small sample size.”); Chen-Oster v. Goldman, Sachs & Co., 114 F. Supp. 3d 110, 120 (S.D.N.Y. 2015) (alteration in original) (“The difficulty with analyzing data on the business unit level, as Goldman Sachs advocates, is that such disaggregation tends to mask common mechanisms because the sample size in each unit is so small. As a general matter, ‘[statistical] power increases with sample size.’”). 16 expressly targeting sex discrimination. Employees also undergo Equal Employment Opportunity training focusing on sexual harassment and bias. Since 2010, KPMG has used a Key Performance Indicator web tool to track diversity data, “monitor the firm’s progress” against diversity goals and “respond as necessary to achieve” those goals. Thirteen female KPMG employees submitted declarations, stating that the Class allegations do not reflect reality; for example, one female partner states: I read the allegations and description of KPMG in the plaintiffs’ Fourth Amended Complaint, and it does not resemble the firm that I know or reflect my own experience at KPMG. I have always been treated with respect, treated as an equal, promoted ahead of my peers, and repeatedly acknowledged based on my performance . . . . In fact, I believe the firm has gone above and beyond in its support . . . . These declarations describe a workplace culture in which women are comfortable reporting instances of misconduct. Defendant also submitted evidence of “Inclusion and Diversity” training conducted at the firm. In contrast, Plaintiffs submitted 150 internal complaints filed by or on behalf of women in Class Positions with KPMG’s Human Resources and Ethics & Compliance departments from 2008 to 2016, including 147 complaints of sexual harassment, “unprofessional behavior” and gender discrimination by or on behalf of 134 women in Class Positions. These internal complaints reflect, among other things, allegations of male coworkers engaging in: rape, attempted rape, sexual assault or groping, unwanted sexual advances or lewd remarks and teamwide visits to strip clubs. Although KPMG terminated at least 12 employees as a result of investigations into misconduct during the Class Period, even after investigating complaints of non-consensual physical contact and finding them substantiated, KPMG sometimes sanctioned the perpetrator with only a reprimand. 17 Plaintiffs also submitted the declarations and deposition testimony of 20 Class Representatives who, in addition to describing conduct similar to that in the HR complaints, also describe a culture where women expect no changes and fear retaliation if they report misconduct.10 Some of the HR Complaints also express mistrust of KPMG’s investigatory process and a fear of retaliation. A 2014 study conducted by KPMG’s Women’s Advisory Board similarly suggested that only 75% of female employees felt they could “report unethical practices without fear of reprisal.” II. DISCUSSION Plaintiffs seek to certify a nationwide class of female Associates, Senior Associates, Managers, Senior Managers/Directors and Managing Directors employed within KPMG’s Tax and Advisory Functions from October 30, 2009, through the date of judgment. Class certification of a Title VII class is denied because Plaintiffs have not shown a common question sufficient to meet the Rule 23 standard enunciated in Dukes, 564 U.S. at 359. Plaintiff Kassman also seeks to certify a class of the same group of employees employed by KPMG in the state of New York from June 2, 2008, through the date of judgment.11 Class 10 Both parties were limited to 30 declarations with respect to the motions for class and final collective certification. 11 Plaintiffs do not seek to certify a New York City subclass, although they assert a claim under the New York City Human Rights Law (the “NYCHRL”). However, they cannot bring claims under the NYCHRL on a statewide basis. See, e.g., Bahl v. N.Y. Coll. of Osteopathic Med. of N.Y. Inst. of Tech., No. 14 Civ. 4020, 2015 WL 4603210, at *10 (E.D.N.Y. July 28, 2015) (stating that the NYCHRL only applies where the “impact of the adverse action” is felt in New York City); Rohn Padmore, Inc. v. LC Play Inc., 679 F. Supp. 2d 454, 463 (S.D.N.Y. 2010) (“[T]he NYCHRL only applies where the actual impact of the discriminatory conduct or decision is felt within the five boroughs, even if a discriminatory decision is made by an employer’s New York City office.”). 18 certification for the subclass is denied because Kassman provided no evidence specific to a New York class. Plaintiffs also seek second stage collective action certification of the Equal Pay Act claims of 1,112 Opt-In Plaintiffs who have been employed in class positions since March 16, 2009. Final certification of an opt-in collective under the Equal Pay Act is also denied because Plaintiffs have not shown that members of the proposed collective work at a single “establishment” or that they are “similarly situated” to one another. A. Evidentiary Motions Before turning to the merits of Plaintiffs’ certification motions, the following discussion defines the universe of materials considered by this Opinion. The parties have filed various motions to strike as well as Daubert motions with respect to each other’s experts. For the reasons below, these motions are denied. 1. Motions to Strike Lay Testimony a. Federal Rules of Evidence Defendant moves to strike portions of the affidavits submitted by 20 female KPMG employees on the basis that their statements (1) are not supported by personal knowledge in violation of Rule 602; (2) constitute inadmissible hearsay and/or (3) are contradicted by their deposition testimony. Defendant also moves to strike paragraph 27 of Attorney Tiseme Zegeye’s Declaration as improper expert testimony. Plaintiffs move to strike portions of KPMG’s memorandum in opposition to class certification that rely on Dr. Bloom’s reiteration of hearsay as facts. The Second Circuit has yet to rule on the standard for motions to strike lay witness evidence at the class certification stage, but “most district courts addressing this question have 19 held that evidence need not be admissible under the Federal Rules of Evidence -- or that the rules should not be applied strictly -- on a motion for class certification.” Flores v. Anjost Corp., 284 F.R.D. 112, 124 n.3 (S.D.N.Y. 2012) (collecting cases); accord Sali v. Corona Reg’l Med. Ctr., 889 F.3d 623, 632 (9th Cir. 2018); but see Unger v. Amedisys Inc., 401 F.3d 316, 319 (5th Cir. 2005) (“[F]indings must be made based on adequate admissible evidence to justify class certification.”). This authority is persuasive. The parties’ arguments concerning testimony challenged under the Federal Rules of Evidence are considered only to evaluate the weight of the evidence and not its admissibility. b. Failures to Disclose Plaintiffs move to strike the organizational chart attached to the declaration of Lynne Doughtie, KPMG’s CEO, on the basis that the chart was not produced during discovery. The motion is denied. Defendant’s counsel represents that the chart was created in conjunction with Lynne Doughtie’s April 17, 2018, declaration -- i.e., after fact discovery was closed. As the chart was created -- based on documents and data previously produced and Doughtie’s knowledge of KPMG’s structure -- to serve as a demonstrative aid in understanding her declaration, it is properly considered in this manner; Plaintiffs are free to attack the chart as misleading, but such attacks go to its weight and not admissibility. See Chen-Oster v. Goldman, Sachs & Co., 114 F. Supp. 3d 110, 130 (S.D.N.Y. 2015) (denying a motion to strike a Goldman Sachs organizational chart not produced in discovery on the basis that it was a demonstrative aid). The motion to strike the organizational chart is denied. Plaintiffs also move to strike portions of the declaration of Joseph Kovatch on the basis that the information underlying the declaration was not produced in discovery. This motion is denied as moot, because the Kovatch Declaration is not relied upon in this Opinion. See, e.g., 20 Patel v. L-3 Commc’ns Holdings Inc., 2016 WL 1629325, at *18 (S.D.N.Y. Apr. 21, 2016) (“Because the Court did not consider Exhibit U in deciding Defendants’ Motion to Dismiss, Lead Plaintiffs’ Motion to Strike is denied as moot.”). c. Violations of Court Orders Plaintiffs move to strike two exhibits as violating prior orders. First, Plaintiffs move to strike a supplemental declaration filed by Dr. Bloom on January 16, 2018, arguing that it violates the deadlines for the production of expert reports and close of expert discovery. Second, Plaintiffs move to strike a 71-page document containing excerpts from 109 resumes in Defendant’s internal personnel database, arguing that it violates a prior order prohibiting excerpts. Likewise, Defendant moves to strike paragraph 4 of Attorney Kate Mueting’s Declaration. Because this Opinion does not rely on any of these materials, the motions are denied as moot. See, e.g., Yearwood v. N.Y. & Presbyterian Hosp., No. 12 Civ. 6985, 2014 WL 1651942, at *13 (S.D.N.Y. Apr. 23, 2014) (“Because the Court has not relied on any other document or representation in the supplemental affidavit, the Hospital’s motion to strike the remainder is denied as moot.”). 2. Daubert Motions The admissibility of expert testimony is governed by Federal Rule of Evidence 702. Pursuant to that Rule: A witness who is qualified as an expert by knowledge, skill, experience, training, or education may testify in the form of an opinion or otherwise if[] (a) the expert’s scientific, technical, or other specialized knowledge will help the trier of fact to understand the evidence or to determine a fact in issue; (b) the testimony is based on sufficient facts or data; (c) the testimony is the product of reliable principles and methods; and (d) the expert has reliably applied the principles and methods to the facts of the case. 21 Fed. R. Evid. 702. “While the proponent of expert testimony has the burden of establishing by a preponderance of the evidence that the admissibility requirements of Rule 702 are satisfied, the district court is the ultimate ‘gatekeeper.’” United States v. Williams, 506 F.3d 151, 160 (2d Cir. 2007) (quoting Daubert v. Merrell Dow Pharm., Inc., 509 U.S. 579, 593 n.10 (1993)); accord In re Pfizer Inc. Sec. Litig., 819 F.3d 642, 658 (2d Cir. 2016). “The district court has broad discretion to carry out this gatekeeping function. Its inquiry is necessarily a ‘flexible one,’ and the types of factors that are appropriate to consider will ‘depend upon the particular circumstances of the particular case at issue.’” Pfizer, 819 F.3d at 658 (first quoting Daubert, 509 U.S. at 594; and then quoting Kumho Tire Co. v. Carmichael, 526 U.S. 137, 150 (1999)). “In deciding whether a step in an expert’s analysis is unreliable, the district court should undertake a rigorous examination of the facts on which the expert relies, the method by which the expert draws an opinion from those facts, and how the expert applies the facts and methods to the case at hand. A minor flaw in an expert’s reasoning or a slight modification of an otherwise reliable method will not render an expert’s opinion per se inadmissible.” Amorgianos v. Nat’l R.R. Passenger Corp., 303 F.3d 256, 267 (2d Cir. 2002); accord United States v. Morgan, 675 F. App’x 53, 55 (2d Cir. 2017) (summary order). “The judge should only exclude the evidence if the flaw is large enough that the expert lacks ‘good grounds’ for his or her conclusions.” Amorgianos, 303 F.3d at 267; accord United States v. Jones, No. 15 Crim. 153, 2018 WL 2684101, at *7 (S.D.N.Y. June 5, 2018). The Supreme Court has outlined four relevant factors for assessing an expert’s reliability: (1) whether a theory or technique can be (and has been) tested; (2) whether the theory or technique has been subjected to peer review and publication; (3) a technique’s known or potential rate of error, and the existence and maintenance of standards controlling the technique’s operation; and (4) whether a particular 22 technique or theory has gained general acceptance in the relevant scientific community. Amorgianos, 303 F.3d at 266 (internal quotation marks omitted) (quoting Daubert, 509 U.S. at 593). These factors “do not constitute a ‘definitive checklist or test . . . . Rather, . . . the trial judge must have considerable leeway in deciding in a particular case how to go about determining whether particular expert testimony is reliable.” Kumho Tire Co., 526 U.S. at 150–52; accord Jones, 2018 WL 2684101, at *7. Although “[t]he Supreme Court has not definitively ruled on the extent to which a district court must undertake a Daubert analysis at the class certification stage,” it has “offered limited dicta suggesting that a Daubert analysis may be required at least in some circumstances.” In re U.S. Foodservice Inc. Pricing Litig., 729 F.3d 108, 129 (2d Cir. 2013); see also Royal Park Invs. SA/NV v. U.S. Bank Nat’l Ass’n, 324 F. Supp. 3d 387, 393 (S.D.N.Y. 2018) (“[T]he Second Circuit has not resolved whether and to what extent Daubert applies at the class certification stage.”). Because of this dicta, and because the parties assume that Daubert applies, this Opinion applies Daubert. See Royal Park Invs., 324 F. Supp. 3d 393. For the reasons below, the motions are denied. a. Dr. Banks Plaintiffs move to exclude Dr. Banks’s report under the first and third factors of Rule 702, arguing that her methodology is unreliable and her opinion is unhelpful. Fed. R. Evid. 702(a), (c). Neither ground supports exclusion pursuant to Daubert. i. Reliability Plaintiffs argue that Dr. Banks’s report is unreliable because she “did not perform a standard job analysis;” rather, she performed an “individualized job analysis,” which Dr. Banks 23 advertises and developed for the purpose of defeating class and collective certification and which lacks peer review. This argument fails. While it is necessary to be skeptical where “experts lacked pre-litigation expertise in the [subject matter] . . . and developed their theories for the purposes of this litigation,” In re Mirena IUD Prods. Liab. Litig., 713 F. App’x. 11, 15 (2d Cir. 2017) (summary order) (citing Washburn v. Merck & Co., 213 F.3d 627 (2d Cir. 2000) (unpublished)), that is not the case here. Dr. Banks was an acknowledged expert in the field of industrial psychology prior to this litigation, whose methodology has gained acceptance in the field; she has been a Senior Lecturer at Berkeley since 1992, and was awarded the Presidential Citation for Innovative Practice by the American Psychological Association in 2009 for her “groundbreaking methodologies” in job analysis. Dr. Banks has published peer reviewed works and given peer reviewed master tutorials to the Society of Industrial and Organizational Psychology on her “individualized job analysis” methodology. Plaintiffs’ depiction of Dr. Banks as advertising her methodology to corporate defendants for the purpose of defeating class certification is a stretch; her PowerPoint states, “I-O psychologists can aid both parties in determining class certifications by defining and distinguishing between and among certain job titles” and “help employees . . . gain the compensation and benefits they earned.” Dr. Banks’s methodology has never been disqualified, despite her testifying as an expert in 83 cases. Plaintiffs further argue that, even if Dr. Banks’s use of an “individualized job analyses” is not totally disqualifying, her opinion should nevertheless be disqualified, because her application of the methodology was unsound for: (1) failing to create a task list; (2) using open ended questions; (3) comparing responses across job title and (4) reporting only dissimilarities in responses while ignoring commonalities. None of these arguments are persuasive. First, to support the argument that failure to create a task list makes a job analysis unreliable, Plaintiffs 24 cite only the report of their own expert, who in turn cites no authority for that proposition. Second, the use of open ended, rather than close ended, questions in a job analysis comes with both advantages and disadvantages, see generally Reference Guide on Survey Research, Federal Judicial Center 391–94 (3d. ed. 2011), and does not necessarily make a study unreliable. See also Guardians Ass’n of N.Y.C. Police Dep’t, Inc. v. Civil Serv. Comm’n of City of N.Y., 630 F.2d 79, 95 (2d Cir. 1980) (approving the use of open ended questions where “the work behaviors involved in being a police officer were identified by extensive interviewing, and subjected to serious review”); Gulino v. Bd. of Educ. of City Sch. Dist. of City of N.Y., 113 F. Supp. 3d 663, 678 (S.D.N.Y. 2015) (excluding a study for failing to use “an open ended investigation into [] job tasks,” because the close ended approach was “inherently flawed”). Third, in contrast to Plaintiffs’ depiction, the majority of Dr. Banks’s report focuses on “comparison of work performed by function and employee level.” Fourth, Dr. Banks reported both similarities and dissimilarities with respect to each question that she coded in frequency tables, stating what percentage of interviewees responded in the same way to a particular question, but determined that only 6 out of 68 questions “showed substantial similarity across interviewees.” Accordingly, Plaintiffs’ motion to exclude Dr. Banks’s report as unreliable is denied. To the extent that Plaintiffs’ arguments raise methodological concerns about Dr. Banks’s reports, those concerns go to weight and not admissibility. See, e.g., Restivo v. Hessemann, 846 F.3d 547, 577 (2d Cir. 2017) (“Frequently, though, ‘gaps or inconsistencies in the reasoning leading to the expert’s opinion . . . go to the weight of the evidence, not to its admissibility.’”) (alteration omitted); see also In re Namenda Direct Purchaser Antitrust Litig., No. 15 Civ. 7488, 2018 WL 3970674, at *11 (S.D.N.Y. Aug. 16, 2018). 25 ii. Helpfulness Plaintiffs also argue that Dr. Banks’s report should be excluded as unhelpful for two reasons. First, Dr. Banks “admitted to not conducting any analysis of the interview responses -statistical or otherwise -- but said she simply reviewed the information provided by the interviewees and reported it to the Court.” Second, “to the extent Dr. Banks performed any qualitative analysis, she did so based not on any specialized knowledge or expertise but only on a reasonable person standard.” The first argument mischaracterizes Dr. Banks’s report and her testimony. Dr. Banks’s report describes her methodology in detail, explaining how she used her expertise to build a study sample, devise the interview questions, conduct interviews and code the responses. Even the portion of Dr. Banks’s deposition that Plaintiffs cite illustrates that she analyzed the interview responses in order to code them: My job is to describe what they do the best way I know how and the most -- in the most reliable and valid manner, and to give this analysis the best opportunity to be judged objectively. So I -- I do the study. I let the facts speak for themselves, and it’s up to the court whether they would draw the same conclusion. . . . [I]f I did my job right in conducting this study and gave what I consider a competent analysis following established practices, then I can draw conclusions on the basis of what I see. (emphasis added). Common sense also suggests that Dr. Banks performed an analysis when conducting her “individualized job analysis.” Plaintiffs’ “reasonable person” argument fails because it too mischaracterizes Dr. Banks’s report and testimony. The content analysis procedure Dr. Banks and her staff used to code the responses was far more complex than a “reasonable person” estimation; it included: (1) devising an initial unique coding scheme for each question; (2) testing the adequacy of the initial coding scheme by coding a sample of data; (3) modifying the initial scheme so it was sufficiently 26 specific to be useful in the analysis but sufficiently broad to cover a majority of responses; (4) calibrating the code by asking interviewers to code randomly selected questions to ensure that they interpreted and applied codes consistently; (5) randomly assigning teams of two to code responses; (6) asking the two coders to reach a consensus if they disagreed and (7) asking a third coder to determine the final code where consensus was not reached. Furthermore, the single statement made by Dr. Banks at her deposition about using a “reasonable person” standard was in response to a question about a hypothetical exercise, not her actual methodology -- i.e., whether “reviewing deliverables and assessing if it is ready to go to the client” is similar work to “producing reports to a client.” Her response was, “It sounds like it’s similar work” based on “[w]hat a reasonable person listening to those two descriptions might say . . . .” Dr. Banks did not say that she used a reasonable person standard when conduction her study. Accordingly, Plaintiffs’ Daubert motion as to Dr. Banks is denied. b. Dr. Vekker Defendant moves to exclude the Report of Dr. Vekker under the first Rule 702 factor, arguing that his opinions are “not connected to the [Equal Pay Act (“EPA”)] claim” and “irrelevant” to the Rule 23 commonality inquiry, because he aggregated data across function. See Fed. R. Evid. 702(a) (requiring that expert testimony “help the trier of fact to understand the evidence or to determine a fact in issue”). In order to determine relevancy under Daubert, the “trial court should look to the standards of Rule 401 in analyzing whether proffered expert testimony is relevant, i.e., whether it ‘ha[s] any tendency to make the existence of any fact that is of consequence to the determination of the action more probable or less probable than it would be without the evidence.’” Amorgianos, 303 F.3d at 265; accord Tramontane v. Home Depot 27 U.S.A., Inc., No. 15 Civ. 8528, 2018 WL 4572254, at *5 (S.D.N.Y. Sept. 24, 2018). Dr. Vekker’s testimony is relevant under this standard. Included within the analyses of Plaintiffs’ disparate treatment and EPA claims are two questions: (1) as a descriptive matter, whether women at KPMG are paid less than men (i.e., correlation), and (2) as a causative matter, whether women at KPMG are paid less than men because of their sex (i.e., causation). See Zarda v. Altitude Express, Inc., 883 F.3d 100, 118 (2d Cir. 2018) (en banc) (“A plaintiff alleging disparate treatment based on sex in violation of Title VII must show two things: (1) that he was ‘discriminated against with respect to his compensation, terms, conditions, or privileges of employment,’ and (2) that the employer discriminated ‘because of sex.’”) (alterations omitted) (quoting 42 U.S.C. § 2000e-2(a)(1)). Likewise, Plaintiffs’ disparate impact claim contains both a correlation and causation element. See Chin v. Port Authority of N.Y. & N.J., 685 F.3d 135, 151 (2d Cir. 2012). At the class certification stage, the key question is whether those mechanisms are common across the class. See Dukes, 564 U.S. at 352 (stating that in most Title VII class actions “proof of commonality necessarily overlaps with [the] merits contention that [the defendant] engages in a pattern or practice of discrimination” because the plaintiffs must be able to “produce a common answer to the crucial question why was I disfavored”) (emphasis in original). Dr. Vekker analyzes only correlation and is up front about that goal: “Dr. Bloom’s analysis fundamentally ignores the relevant question that I am answering . . . the question in this case whether there are gender disparities in compensation and selections for promotions in Tax and Advisory functions at KPMG.” Dr. Vekker’s regressions convincingly answer this threshold question of correlation in the affirmative, which is an important starting point. Defendant is correct that, for the reasons discussed below, Dr. Vekker’s report, coupled with the remainder of 28 Plaintiffs’ evidence, does not convincingly show that causation can be determined on a classwide basis -- but this does not mean that Dr. Vekker’s testimony is unhelpful. Dr. Vekker’s testimony need only logically advance Plaintiffs’ case. It does so by showing that disparities exist and affect all members of the Proposed Class. The motion is denied. c. Dr. Goldberg Defendant moves to exclude the testimony of Plaintiffs’ Human Resources expert, Dr. Caren Goldberg, arguing that her opinion that KPMG policies do not comply with best practices is irrelevant because she fails to demonstrate that the policies caused women to be paid less. This argument fails. As with Dr. Vekker, Dr. Golberg’s failure to address causation does not make her opinion irrelevant, as no single expert must prove a party’s entire case. See Moussouris, 311 F. Supp. 3d at 1237 (“It is certainly true that Dr. Farber did not draw any specific casual links between the Calibration Process and the disparities he found. But that does not render Dr. Farber’s analysis of the alleged disparity irrelevant to the question of commonality.”) (citation omitted); Chen-Oster, 114 F. Supp. 3d at 125 (“[I]t is not necessary for each expert to provide evidence establishing every element of a party’s case . . . .”). Dr. Goldberg merely needs to advance Plaintiffs’ case, which she does by suggesting KPMG’s compensation and promotion policies, which operated uniformly across the class, do not comply with HR best practices and are not justified by business necessity. See Chen-Oster, 114 F. Supp. 3d at 126 (“[E]xpert evidence that addresses the reliability of Goldman Sachs’ evaluation processes goes to the business necessity defense and is relevant at the class certification stage to the question of commonality.”); Houser v. Pritzker, 28 F.Supp.3d 222, 243 (S.D.N.Y. 2014) (holding commonality established by defenses applicable to all class members); see also McReynolds v. Merrill Lynch, Pierce, Fenner & Smith, Inc., 672 F.3d 482, 489 (7th Cir. 29 2012) (holding that whether “teaming” policy is justified by business necessity is an issue “common to the entire class and therefore appropriate for class-wide determination”). The motion to exclude Dr. Goldberg is denied. d. Additional Motions Plaintiffs’ motion to exclude Dr. Stockdale and Defendant’s motion to exclude Dr. Hanges are denied as moot, because those experts are not relied upon in this Opinion. See, e.g., Axiom Inv. Advisors, LLC v. Duetsche Bank AG, No. 15 Civ. 9945, 2018 WL 4253152, at *8 n.7 (S.D.N.Y. Sept. 6, 2018) (“Deutsche Bank’s Daubert motion to strike Feller’s expert testimony is denied as moot because the Court does not rely on Feller’s testimony in ruling on Plaintiff’s motion for class certification.”). B. Title VII Class Certification Federal Rule of Civil Procedure 23(a) provides that plaintiffs may sue on behalf of a class where: (1) the class is so numerous that joinder of all members is impracticable; (2) there are questions of law or fact common to the class; (3) the claims or defenses of the representative parties are typical of the claims or defenses of the class; and (4) the representative parties will fairly and adequately protect the interests of the class.12 The Second Circuit “has also ‘recognized an implied requirement of ascertainability in Rule 23,’ which demands that a class be ‘sufficiently definite so that it is administratively feasible for the court to determine whether a particular individual is a member.’” In re Petrobras Secs., 862 F.3d 250, 260 (2d Cir. 2017) (quoting Brecher v. Republic of Argentina, 806 F.3d 22, 24 (2d Cir. 2015)). 12 This Opinion is mindful that Rule 23 will be updated effective December 1, 2018. 30 The Second Circuit gives Rule 23 a “liberal rather than restrictive construction, and courts are to adopt a standard of flexibility.” Marisol A. v. Giuliani, 126 F.3d 372, 377 (2d Cir. 1997); accord In re J.P. Morgan Stable Value Fund ERISA Litig., No. 12 Civ. 2548, 2017 WL 1273963, at *5 (S.D.N.Y. Mar. 31, 2017). But “Rule 23 does not set forth a mere pleading standard.” Dukes, 564 U.S. at 350. Plaintiffs must establish by a preponderance of the evidence that each of Rule 23’s requirements is met. In re Vivendi, S.A. Secs. Litig., 838 F.3d 223, 264 (2d Cir. 2016). A certifying court “must receive enough evidence, by affidavits, documents, or testimony, to be satisfied that each Rule 23 requirement has been met.” Shahriar, 659 F.3d at 251 (citing In re IPO Secs. Litig., 471 F.3d 24, 41 (2d Cir. 2006)). Although “a court’s class-certification analysis must be ‘rigorous’ and may ‘entail some overlap with the merits of the plaintiff’s underlying claim,’ Rule 23 grants courts no license to engage in free-ranging merits inquiries at the certification stage.” Amgen Inc. v. Conn. Ret. Plans & Tr. Funds, 568 U.S. 455, 465–66 (2013) (citing Dukes, 564 U.S. at 351). Any factual disputes relevant to satisfying each Rule 23 requirement should be resolved, but a court “should not assess any aspect of the merits unrelated to a Rule 23 requirement.” In re IPO Secs. Litig., 471 F.3d at 41. As is true in many Title VII class actions, the “crux of this case is commonality.” See Dukes, 564 U.S. at 349. Rule 23(a)(2) requires a plaintiff to show that “there are questions of law or fact common to the class.” Fed. R. Civ. P. 23(a)(2). “[E]ven a single common question will do” to satisfy the commonality requirement of Rule 23(a)(2). Dukes, 546 U.S. at 359 (alterations and internal quotation marks omitted). However, because any competently crafted class complaint “literally raises common ‘questions,’” not every common question suffices. Id. at 349. Instead, Plaintiffs’ claims must “depend upon a common contention” that is “of such a nature that it is capable of classwide resolution -- which means that determination of its truth or falsity will 31 resolve an issue that is central to the validity of each one of the claims in one stroke.” Id. at 350. In other words, Plaintiffs must pose a common question that “will produce a common answer to the crucial question why was I disfavored.” Id. at 352. Certain requirements for class certification, including the elements of commonality, typicality and predominance, can be assessed only in light of the substantive law governing a case. The core provision of Title VII states that: It shall be an unlawful employment practice for an employer . . . to fail or refuse to hire or to discharge any individual, 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 . . . . 42 U.S.C. § 2000e-2(a)(1). Plaintiffs here assert claims of both disparate impact and disparate treatment under Title VII. Plaintiffs’ motion to certify a class with respect to either or both claims is denied for failure to satisfy the commonality requirement, for the reasons explained below. 1. Plaintiffs Cannot Show Commonality on Their Disparate Impact Claim a. Governing Law Under a disparate impact theory of Title VII liability, a plaintiff may prevail by demonstrating that the employer “uses a particular employment practice that causes a disparate impact on the basis of race, color, religion, sex, or national origin.” 42 U.S.C. § 2000e2(k)(1)(A)(i). Title VII prohibits some employment practices “that are not intended to discriminate but in fact have a disproportionately adverse effect” on a protected class. Ricci v. DeStefano, 557 U.S. 557, 577 (2009). To prove disparate impact discrimination, a plaintiff must “(1) identify a specific employment practice or policy; (2) demonstrate that a disparity exists; and 32 (3) establish a causal relationship between the two.” Chin, 685 F.3d at 151 (internal quotation marks and citations omitted). Dukes addressed the issue of commonality in a Title VII gender discrimination case that challenged the discretion exercised by individual supervisors over pay and promotion matters. See 564 U.S. at 344–45. The Dukes plaintiffs alleged that “local managers’ discretion over pay and promotions [was] exercised disproportionately in favor of men, leading to an unlawful disparate impact on female employees.” Id. Plaintiffs’ theory of commonality was that WalMart’s “strong and uniform ‘corporate culture’ permits bias against women to infect . . . the discretionary decisionmaking of each one of Wal-Mart’s thousands of managers -- thereby making every woman at the company the victim of one common discriminatory practice.” Id. at 345. The Supreme Court held that the class could not be certified and observed that “[t]he only corporate policy that the plaintiffs’ evidence convincingly establishes is Wal-Mart’s ‘policy’ of allowing discretion by local supervisors over employment matters.” Id. at 355. Such a policy of discretion is, “[o]n its face . . . just the opposite of a uniform employment practice that would provide the commonality needed for a class action; it is a policy against having uniform employment practices.” Id. Because “demonstrating the invalidity of one manager’s use of discretion will do nothing to demonstrate the invalidity of another’s,” a party basing commonality on such a system of discretion “will be unable to show that all the employees’ Title VII claims will in fact depend on the answers to common questions.” Id. at 355–56. The Court in Dukes recognized that in appropriate cases, granting unlimited discretion to lower-level supervisors could be the basis of Title VII liability. Id. at 355; see Scott v. Family Dollar Stores, Inc., 733 F.3d 105, 113–14 (4th Cir. 2013) (describing how a disparate impact 33 claim challenging discretion may satisfy commonality post-Dukes). Plaintiffs challenging a system of discretion must identify “a common mode of exercising discretion that pervades the entire company.” Dukes, 564 U.S. at 356; see also Chen-Oster v. Goldman, Sachs & Co., 325 F.R.D. 55, 73 (S.D.N.Y. 2018). In other words, when plaintiffs wish to challenge numerous employment decisions at once, they must point to “some glue holding the alleged reasons for all those decisions together.” Dukes, 564 U.S. at 352. Otherwise, it would be “quite unbelievable that all managers would exercise their discretion in a common way without some common direction.” Id. at 356. The Supreme Court determined that both the plaintiffs’ aggregated statistical evidence and anecdotal evidence fell “well short” of indicating a “common direction” from upper management, and thus, the plaintiffs failed to demonstrate the existence of common issues. Id. at 356–58. On remand, the Dukes plaintiffs narrowed their class significantly, from a nationwide class to three regional classes, all within California. Dukes II, 964 F. Supp. 2d at 1118–19. They identified several corporate policies, such as a “tap on the shoulder” system for promotions and company-wide guidelines for pay decisions, which allegedly provided “common direction.” Id. at 1119–25. But the alleged promotion and pay criteria “were so vague or numerous that they imposed no real constraints,” and the defendant did not prohibit managers from considering other factors of their own choosing. Id. at 1126. The plaintiffs were not challenging the policies themselves or the faithful application of those policies; instead, the plaintiffs’ argument still boiled down to the theory that “managers, who were left without meaningful guidance in applying the impossibly vague criteria, fell back on their own stereotyped views of women in making pay and promotion decisions.” Id. at 1127. While the court found this to be “a perfectly logical theory, . . . it leaves [p]laintiffs right back where they started: challenging Wal-Mart’s 34 practice of delegating discretion to local managers, which the Supreme Court specifically held was not a specific employment practice supplying a common question sufficient to certify a class.” Id. As in Dukes II, other courts addressing challenges to discretionary pay and promotion practices have concluded that commonality is lacking. See Tabor v. Hilti, Inc., 703 F.3d 1206, 1229 (10th Cir. 2013) (recognizing that courts “have generally denied certification when allegedly discriminatory policies are highly discretionary”). In Jones v. National Council of Young Men’s Christian Associations, the plaintiffs argued that the company did not have “effective oversight” and thus allowed local managers’ exercise of discretion to result in adverse outcomes for the protected class. 34 F. Supp. 3d 896, 905 (N.D. Ill. 2014). Plaintiffs attempted to “downplay[] the degree of discretion” by describing the entire evaluation and promotion process as a mandatory company-wide policy. Id. at 904. The court in Jones rejected this attempt to “increas[e] the level of abstraction in defining the policy.” Id.; see id. at 905 (observing that at the level of abstraction the plaintiffs propose, “every company -- even WalMart -- could be said to have a company-wide policy”). Instead, the court characterized the plaintiffs’ challenge to oversight as “precisely the argument held to be inadequate to support class certification in [Dukes].” Id. at 905; see also Bolden v. Walsh Constr. Co., 688 F.3d 893, 898 (7th Cir. 2012) (“[Dukes] tells us that local discretion cannot support a company-wide class no matter how cleverly lawyers may try to repackage local variability as uniformity.”). After Dukes, for discretionary pay and promotion procedures to satisfy commonality, lower-level supervisors must operate under “a common mode of exercising discretion that pervades the entire company,” such that individual discretionary decisions nonetheless produce a common answer to the question “why was I disfavored.” See 564 U.S. at 352, 356 (emphasis in 35 original); see also Ross v. Lockheed Martin Corp., 267 F. Supp. 3d 174, 198 (D.D.C. 2017) (requiring the plaintiffs to demonstrate that “all managers would exercise their discretion in a common way”). Factors that are helpful to consider in determining whether a common mode of exercising discretion pervades the entire company include: (1) the nature of the purported class; (2) the process through which discretion is exercised; (3) the criteria governing the discretion and (4) the involvement of upper management. See Moussouris, 2018 WL 3328418, at *16 (collecting cases). b. Application to Plaintiffs’ Claims All four factors suggest that Plaintiffs have not shown that their disparate impact claims depend on a common question whose answer will resolve an issue that is central to the validity of each class member’s claims. See Dukes, 546 U.S. at 350. i. Class Size The nature of the purported class is relevant to the commonality inquiry. Although class size “has no per se bearing on commonality,” courts recognize that “when the claims focus in part on the exercise of managerial discretion, it is reasonable to suspect that the larger the class size, the less plausible it is that a class will be able to demonstrate a common mode of exercising discretion.” See Ellis v. Costco Wholesale Corp., 285 F.R.D. 492, 509 (N.D. Cal. 2012) (citing Chen-Oster v. Goldman, Sachs & Co., 877 F. Supp. 2d 113, 119 (S.D.N.Y. 2012)). “[A] more centralized, circumscribed environment generally increases . . . the consistency with which managerial discretion is exercised . . . .” Brown v. Nucor Corp., 785 F.3d 895, 910 (4th Cir. 2015); see also Rollins v. Traylor Bros., Inc., No. 14 Civ. 1414, 2016 WL 258523, at *7 (W.D. Wash. Jan. 21, 2016), abrogated on other grounds, 2016 WL 5942943 (W.D. Wash. May 3, 2016). In Brown, for example, the proposed class consisted of about 100 putative members, all 36 located in a single facility. 785 F.3d at 910. For a “localized, circumscribed class” of that size, a policy of subjective, discretionary decision-making can “more easily form the basis of Title VII liability.” Id. at 916; see also Chen-Oster, 325 F.R.D. at 62–63 (approving class of 1,762 to 2,300 people holding two job positions). On the other hand, for a nationwide class, “proving a consistent exercise of discretion will be difficult, if not impossible.” Brown, 785 F.3d at 916. Plaintiffs seek to certify a nationwide class of at least 10,000 women in various offices throughout the country. Although Plaintiffs’ class size is a small fraction of the 1.5 million putative class members in Dukes, 564 U.S. at 342, the proposed class here is still several times the size of the certified classes in Ellis, 285 F.R.D. at 509 (700 members), Chen-Oster, 325 F.R.D. at 62–63 (between approximately 1,762 and 2,300 members) and Brown, 785 F.3d at 910 (100 members). Nor is the proposed Nationwide Class centralized or localized. Unlike Ellis -where all class members held “two closely related” positions sharing “a uniform job description across the class,” 285 F.R.D. at 509 -- or Chen-Oster -- where class members held two positions, 325 F.R.D. at 63 -- Plaintiffs’ proposed class covers a myriad of job descriptions, ranging from accountants to scientists, engineers, attorneys and Ph.D. economists, at levels of seniority ranging from Associate to Managing Director. And unlike those in Brown, the proposed class members in this case are not located in a single location. See 785 F.3d at 910. They are located all over the country. Plaintiffs’ proposed class is more akin to the expansive Dukes class than the circumscribed classes in Ellis, Chen-Oster and Brown. ii. Framework for Discretion Second, to determine whether a common mode of exercising discretion pervades the entire company, courts look to what procedures govern the discretion and then analyze the rigidity of the process through which discretion is exercised. A process that constrains and 37 channels the exercise of discretion may show sufficient common direction. In Ellis, the court noted that Costco not only had a single procedure that governed promotion decisions, but the company also imposed several non-negotiable requirements, such as mandatory promotion from within the company and a prohibition on posting job openings. See 285 F.R.D. at 511. Those requirements -- implemented company wide -- distinguished the discretion at issue in Ellis from the unfettered discretion in Dukes. See Ellis, 285 F.R.D. at 509; see also McReynolds, 672 F.3d at 489–90 (holding that two company-wide policies that required the managers to allow certain practices, rather than leaving managers to make the decision themselves, sufficiently restrained discretion). KPMG’s pay and promotion procedures act more as a framework that dictates who will make discretionary decisions rather than how they will exercise their discretion. Broadly speaking, putative class members were all subject to “uniform and firm-wide compensation and promotion practices.” Employees participate in goal-setting meetings with their PMLs, and are evaluated in practice area assessment meetings at which eligible employees are reviewed for promotion. Under the direction of Compensation Strategies, the company sets salary ranges and recommended pay increases and bonuses for employees depending on, among other factors, their specific position and market rates for that position, current salary and performance. The subpractice leader in charge of compensation for a practice area decides the compensation of the employees in that practice area. These decisions are reviewed in the aggregate for compliance with the budget by higher management. These procedures, while uniform and firm-wide, merely “set[] up a structure” through which employees are evaluated. Valerino v. Holder, 283 F.R.D. 302, 313 (E.D. Va. 2012). Like the evaluation and promotion process in Valerino, the process Plaintiffs identify reveals only stages or phases, during which individual decision makers or 38 groups at the practice area level exercised discretion. See id. at 313–14. Identifying the framework says nothing about whether the exercise of discretion itself can “be said to be the same or ‘common.’” Id. at 313. Plaintiffs argue, relying on the testimony of Dr. Goldberg, that common compensation policies exist based upon KPMG’s use of overlapping salary ranges, which “divorce[] pay from performance” to women’s detriment, and its use of “prior salaries” to set current salaries, which “entrenches pre-existing gender disparities.” Neither argument is persuasive. The “overlap” argument fails for two reasons. First, it is possible that overlapping merit and variable compensation pay scales disentangle pay and performance by allowing lower rated men to be compensated better than higher rated women -- e.g., under the 2012 Tax Managing Director Salary Increase Guidelines, a lower 3-rated employee could earn a 5% merit increase while a higher 1-rated employee might earn only a 4% increase. However, Plaintiffs have not provided evidence that a single woman was disadvantaged in this manner, as Dr. Goldberg acknowledged at her deposition, let alone “significant evidence” that “overlap” operated as a common causal mechanism across the class. Second, the “overlap” argument boils down to a claim that discretion delegated to lower-level managers allowed bias to infect compensation decisions. As described in Dr. Goldberg’s report, overlapping pay scales are problematic because they “create an opaque system that provides fertile ground for bias.” Plaintiffs’ “compounding pay gap” argument also fails, because they have not shown an initial pay gap.13 In support of this argument, Plaintiffs cite the Ninth Circuit’s recent holding 13 This argument is also properly rejected because Plaintiffs raised it in one conclusory sentence in their opening memorandum of law and developed it only in their reply. See, e.g., Jaen v. Sessions, 899 F.3d 182, 187 (2d Cir. 2018) (quoting Norton v. Sam’s Club, 145 F. 3d 114, 117 39 “that prior salary alone or in combination with other factors cannot justify a wage differential.” Rizo v. Yovino, 887 F.3d 453, 456 (9th Cir. 2018). However, Rizo was not a class action. Id. at 457–58. In that case, one female plaintiff was paid less at her new job than men holding the same position, because she had been paid less at her previous job than her male counterparts had been paid at their previous jobs. Id. at 458–59. But in this case, Plaintiffs have not shown that their source salaries -- i.e., their original salaries at KPMG -- were lower than the source salaries of their male counterparts. Accordingly, Plaintiffs do not provide “significant proof” of a common initial pay disparity that KPMG’s compensation system could have compounded over time. To the extent there is relevant evidence in the record, it cuts against the “compounding pay gap” theory, because, as Dr. Goldberg puts it, KPMG “moves” employees “to the middle” of their pay bands over time by “putting more money towards groups that are further behind [in] the compensation tool.” While this policy of moving people “to the middle” might “decouple[] pay and performance,” it also presumably dissipates any initial sex-based pay discrepancies over time. Plaintiffs also argue that KPMG uses a common “tap on the shoulder” system for promotions to the detriment of women, because it does not post job openings and invite applications. This mischaracterizes how promotions are made at KPMG. A “tap on the shoulder” system implies a setting where promotions are scarce and made at irregular intervals, so that only employees who are “tapped on the shoulder” are promoted or know to apply. See Dukes, 564 U.S. 371 (Ginsburg, J., dissenting). However, the record reflects that KPMG employees operate on a career track; they are routinely reviewed for promotion when they meet (2d Cir. 1998)) (“Issues not sufficiently argued in the briefs are considered waived . . . .”); accord Pirnik v. Fiat Chrysler Autos., N.V., 327 F.R.D. 38, 44 n.2 (S.D.N.Y. 2018). 40 certain eligibility requirements, and therefore need not apply or be “tapped on the shoulder.” PMLs and other practice area members discuss an employee’s promotion readiness with the employee as a matter of course at several points throughout the performance management process, including (1) at the annual goal-setting meeting with the PML, (2) prior to mid-year review meetings and (3) prior to the year-end assessment meetings. Promotion readiness is also evaluated for each eligible employee at the year-end practice area assessment meetings. iii. Criteria Governing Discretionary Decisions Third, in assessing whether a discretionary system satisfies commonality, courts consider the criteria that govern the discretionary decisions. Again, whether a set of criteria creates a common mode of exercising discretion depends on the rigidity of the criteria. Subjective criteria, prone to different interpretations, generally do not provide common direction. See, e.g., Ross, 267 F. Supp. 3d at 198 (“[T]hat all . . . supervisors used the same allegedly ill-defined numerical rubric . . . says nothing about how individual supervisors exercised what discretion was left to them.”); Jones, 34 F. Supp. 3d at 906 (“That supervisors evaluate candidates according to specific, but subjective, factors . . . does not make the decisions produced by the process meaningfully less discretionary.”) But objective criteria, or even subjective criteria defined uniformly, may establish a common mode of exercising discretion. In Ellis, the court found commonality sufficient for class certification in part because most of the promotion criteria -such as the applicant’s ability to relocate or an applicant’s prior experience in the position -- were objective and thus not open to interpretation by individual decision-makers. See 285 F.R.D. at 514, 531. The remaining criteria, while subjective, were uniformly interpreted within the company. See id. at 518–19. 41 KPMG’s evaluation and promotion criteria are not sufficiently specific to constrain discretion. Plaintiffs argue that KPMG’s numerical evaluation metrics are “vague and unweighted” like the “allegedly ill-defined numerical rubric” in Ross, 267 F. Supp. 3d at 198, and that many of KPMG’s promotions criteria are abstract like the “amorphous concepts” in Jones, 34 F. Supp. 3d at 906, including considerations like “professionalism,” “integrity,” “reputation” and potential to be a “partner candidate.” Plaintiffs do not present evidence that any of these challenged criteria constrained discretion. See id. at 905–06. While this “delegation of discretion over pay and promotions is a policy uniform throughout” KPMG, Dukes, 564 U.S. at 377 (Ginsburg, J., dissenting) -- and, as explained by Dr. Goldberg, the use of nebulous criteria such as “partner candidate” by managers at KPMG very likely disadvantages women in a profession traditionally dominated by men -- it is nevertheless the type of uniform policy that Dukes held cannot support a finding of commonality. 564 U.S. at 356–57. iv. Managerial Non-involvement Fourth, and last, the involvement of top management in the discretionary decision-making is a key consideration in assessing commonality. See Dukes, 564 U.S. at 350 (suggesting that commonality may be satisfied if class members were subject to the bias of a single supervisor); Brown, 785 F.3d at 916 (recognizing commonality if high-level personnel exercise the discretion at issue). If, for instance, “a single decisionmaker[] or a small, cohesive group” vetted all of the pay and promotion decisions, then the involvement of that one individual, or small group of individuals, could constitute a common practice. Jones, 34 F. Supp. 3d at 908; see Chen-Oster, 325 F.R.D. at 66, 76, 84 (certifying a class where upper management developed the list of eligible candidates and “[u]ltimately, [the] management committee decide[d] who [wa]s promoted”); Ellis 285 F.R.D. at 511–12 (finding commonality where, even though lower-level managers had input, 42 top management actually made the promotion decisions). But mere approval or limited oversight by higher-level executives, without more, falls short of showing a sufficient common denominator. See Jones, 34 F. Supp. 3d at 908. Plaintiffs have not shown sufficient involvement by top management. Plaintiffs claim that “ultimately all pay and promotion decisions must be approved by two individuals: the Vice Chair and National Managing Partner for Tax and Advisory.” But final approval in the aggregate does not show commonality in the exercise of discretion. As in Jones, Plaintiffs provide no evidence that levels of management above the more than 150 practice area leaders reviewed individual promotion recommendations with “any frequency, much less that they did so with . . . regularity.” 34 F. Supp. 3d at 908. In the absence of contradictory evidence, the Court credits the sworn statement of Lynne Doughtie, KPMG’s CEO and former National Managing Partner and then Vice Chair of Advisory, that members of senior management “do not review or secondguess individual employee performance, pay or promotion decisions . . . .” Rather, upper management “simply confirm[s] that, as a matter of process, assessment meetings were held and the compensation process completed” and “look[s] at aggregated pay and promotion decisions against budgets.” The involvement of KPMG senior management in pay and promotion decisions more closely resembles the “limited oversight” in Jones, 34 F. Supp. 3d at 908, than the “consistent[] and pervasive” involvement in Ellis, 285 F.R.D. at 511–12. In sum, these factors -- the class size of thousands of female employees, the relative freedom and independence extended to local supervisors who made pay and promotion decisions, the lack of rigid criteria to channel discretion and the lack of review by senior management of individual personnel decisions -- all dictate the conclusion that there was no common mode of exercising discretion that would warrant class certification after Dukes. See Dukes II, 964 F. 43 Supp. 2d at 1127. Accordingly, Plaintiffs motion for class certification of a disparate impact class is denied. 2. Plaintiffs Cannot Show Commonality on Their Disparate Treatment Claim Unlike disparate impact claims, where the employer’s practice or policy is facially neutral, a disparate treatment claim is based on the allegation that the defendant engaged in intentional discrimination. The plaintiff must present “evidence supporting a rebuttable presumption that an employer acted with the deliberate purpose and intent of discrimination against an entire class.” United States v. City of New York, 717 F.3d 72, 87 (2d Cir. 2013) (“City of New York”); accord EEOC v. Mavis Discount Tire, Inc., 129 F. Supp. 3d 90, 103 (S.D.N.Y. 2015). Disparate treatment claims do not require plaintiffs to identify a specific company-wide employment practice responsible for the discrimination. See Int’l Bhd. of Teamsters v. United States, 431 U.S. 324, 336 n.16 (1977). In a disparate treatment class action, plaintiffs must provide evidence of a “systemwide pattern or practice” of pervasive discrimination against the class, such that the discrimination is “the regular rather than the unusual practice.” Id. at 336. Such a showing creates a rebuttable presumption of the required discriminatory intent. City of New York, 717 F.3d at 83. Plaintiffs may satisfy their burden of showing a pattern or practice of discrimination through statistics alone. Teamsters, at 339–40; see Hazelwood School Dist., 433 U.S. at 307–08 (“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.”). Although anecdotal evidence of discrimination is relevant to show a policy of intentional discrimination, such evidence is not required. City of New York, 717 F.3d at 84. 44 If plaintiffs satisfy their burden, the burden of production then shifts to the employer to show a non-discriminatory reason for the apparently discriminatory result. City of New York, 717 F.3d at 84–85. The employer may rebut the inference of discriminatory intent by offering any relevant evidence. Id. at 85–87. The ultimate burden of persuasion remains with the plaintiff to show that the employer acted with discriminatory intent. Id. at 87. In a disparate treatment case, proof of commonality for purposes of Rule 23 overlaps with Plaintiffs’ prima facie burden to show that KPMG engaged in a systemwide pattern or practice of discrimination. See Dukes, 564 U.S. at 352. There is a “wide gap” between an individual’s claim that he was subjected to a company’s policy of discrimination and the existence of a class of persons who have suffered the same injury as that individual such that the individual’s claim and the class claims will share common questions of law or fact. Id. at 352–53 (quoting Gen. Tel. Co. of Sw. v. Falcon, 457 U.S. 147, 157 (1982)). To bridge that gap, Plaintiffs must provide “significant proof” that KPMG “operated under a general policy of discrimination.” Id. at 353 (quoting Falcon, 457 U.S. at 159, n.15). Here, Plaintiffs point to “statistically significant disparities resulting from common compensation and promotion processes; . . . evidence of knowledge on the part of KPMG’s senior leadership [and] a failure of KPMG’s senior leadership to remedy known disparities; and . . . substantial anecdotal evidence.” This evidence is insufficient to show, for purposes of Rule 23 commonality, that KPMG operated under a general policy of discrimination. a. Statistical Evidence Contrary to their assertion, Plaintiffs’ evidence does not show statistically significant disparities between similarly situated men and women in pay and promotion -- much less the kind 45 of gross disparities that on their face would suggest discriminatory intent. See Teamsters, 431 U.S. at 339. The plaintiffs in Dukes offered a statistical study similar to that of Plaintiffs’ statistical expert Dr. Vekker, in which regression analyses revealed statistically significant disparities between men and women that “can be explained only by gender discrimination.” 564 U.S. at 356. Dukes held that this statistical evidence was insufficient because it suffered from a “failure of inference.” Id. The evidence established only that there was a disparity at the national or regional level, but information at that level did not establish the existence of disparities at individual stores, let alone “the inference that a company-wide policy of discrimination is implemented by discretionary decisions at the store and district level.” Id. at 356–57. For example, a regional pay disparity could be attributable to a small set of stores, and cannot alone establish a “uniform, store-by-store disparity.” Id. at 357. Thus, these higher-level statistics could not produce a common answer to the question “why was I disfavored.” Id. at 352. Post-Dukes, the statistical study must “conform[] to the level of decision for the challenged practices.” Ellis, 285 F.R.D. at 523. When decisions are made at the discretion of local decision makers, courts may find aggregated statistical evidence inadequate because it is “derived from hundreds of employment decisions made by myriad decision makers, at different times, under mutable procedures and guidelines, in different departments, and in different office locations, concerning employees at varying levels of experience, responsibilities, and education.” Jones, 34 F. Supp. 3d at 909. But if plaintiffs demonstrate that practices are uniform across the company, then courts have found “good reason to rely on nationwide statistics.” Ellis, 285 F.R.D. at 523 (relying on aggregated statistics because the CEO averred that the promotion policies derived from the top management and were uniform across the company). 46 Plaintiffs here have not shown that promotion policies and practices are uniform across KPMG. Thus, there is no “good reason to rely on nationwide statistics.” Id. The relevant level of decision-making for the challenged practices here remains at the practice-area level. Dr. Vekker’s statistical study, however, is based on aggregate figures. Dr. Vekker ran regressions comparing all relevant employees at the Function level (i.e, the Tax Function and the Advisory Function), in line with Plaintiffs’ theory that KPMG’s Function- and firm-wide pay and promotion policies had a disparate impact on women. Even accepting the validity of Dr. Vekker’s statistical analysis, his evidence “has the same problem as the statistical evidence in [Dukes]: it begs the question.” Bolden, 688 F.3d at 896. Dr. Vekker’s finding of disparity “may be attributable to only a small set of [KPMG decision makers at the practice-area level], and cannot by itself establish the uniform . . . disparity upon which the [P]laintiffs’ theory of commonality depends.” See Dukes, 564 U.S. at 357. Furthermore, the regressions of Defendant’s expert Dr. Bloom suggest that Service Line, rather than gender, accounts for any pay and promotion disparities. When Dr. Bloom compared men and women who hold the same job title within the same Service Line, gender disparities largely disappeared, suggesting that Dr. Vekker’s decision to aggregate data across function, thereby equating job and job title, obfuscated the principal explanatory variable. Dr. Vekker’s results are likely explained by a higher concentration of men in the best compensated Service Lines and Cost Centers within a single function -- e.g., Managing Directors in the SALT Income and Incentives Cost Center earn more than three times as Managing Directors in the BTS High Volume Trust Tax Cost Center -- even though men and women with the same job title within the same Service Lines and Cost Centers earn the same amounts. Dr. Vekker’s statistical evidence is insufficient to show any common issue that would warrant a nationwide class. 47 b. Failure to Remedy Known Disparities Plaintiffs claim that KPMG was aware that its pay and promotion policies resulted in a disparate impact against women and allowed those policies to continue. This argument mischaracterizes the evidence. The documents Plaintiffs cite generally concern KPMG’s diversity efforts, which lay out KPMG’s shortcomings as well as successes with regard to diversity and inclusion, and also steps KPMG is taking to address shortcomings. For example, Plaintiffs cite a study performed by DiversityInc. as standing for the proposition that “KPMG acknowledged as early as 2009 that ‘compensation disparities still exist,’” but the document Plaintiffs cite actually lauds KPMG’s progress: KPMG has dramatically increased its position as a diversity leader through concerted efforts in talent development, mentoring and employee resource groups. In 2008, the firm’s initial placement was 49, then advanced to 21 in 2009. . . . Our efforts with women have strongly contributed to the increase in ranking; however the number of women on the Board will adversely impact us in the future. . . . The firm should focus on increased diversity of senior management. While we have done a good job with developing and retaining women, compensation disparities still exist. Plaintiffs claim that KPMG failed to “review or analyze the centrally determined compensation and promotion decisions to see if women are paid or promoted equally.” This argument is unavailing, first, because (as discussed above) compensation and promotion decisions are not “centrally determined,” but rather, left to the discretion of practice area leaders within general guidelines set by the firm. Second, in response to the 2009 DiversityInc. report’s recommendation that KPMG “[a]nalyze compensation data for women and minorities across all levels,” KPMG began doing so in 2010 via its Key Performance Indicator and a yearly anonymous Work Environment Survey (now called “Employee Engagement Surveys”). 48 Plaintiffs cite a discovery letter from KPMG stating that the company is not aware of analyses of “any alleged gender impact of KPMG’s compensation, promotion or evaluation systems.” This representation is not the same as saying that senior leadership at KPMG was inattentive to the issue. Several documents show that the firm was sensitive to gender disparities in pay and promotion and made efforts to remedy the issue. Through their expert Dr. Goldberg, Plaintiffs claim that KPMG does not conduct training specifically targeted at avoiding bias in compensation and promotion decisions, and that the diversity training KPMG does conduct is ineffective. Plaintiffs’ evidence does not amount to “significant proof” that KPMG “operated under a general policy of discrimination.” Dukes, 564 U.S. at 353. At most, Plaintiffs have shown that KPMG was aware of a pay and promotions gap, and that the firm’s efforts have not completely eradicated the gap. c. Anecdotal Evidence Plaintiffs’ anecdotal evidence also does not constitute the necessary “significant proof” of classwide discriminatory intent. A disparate treatment claim often includes “a combination of strong statistical evidence of disparate impact coupled with anecdotal evidence of the employer’s intent to treat the protected class unequally.” Mozee v. Am. Commercial Marine Serv. Co., 940 F.2d 1036, 1051 (7th Cir. 1991); see also Robinson v. Metro-N. Commuter R.R. Co., 267 F.3d 147, 158 (2d Cir. 2001). In Teamsters, the Supreme Court found sufficient evidence of a systemwide pattern or practice of employment discrimination based on pervasive statistical disparities “bolstered” by anecdotal evidence of discrimination. 431 U.S. at 337–88. Anecdotal evidence by definition is individual rather than widespread; it serves an illustrative purpose to bring “the cold numbers convincingly to life.” Id. at 339. “While anecdotal evidence may suffice to prove individual claims of discrimination, rarely, if ever, can such evidence show a systemic 49 pattern of discrimination.” Middleton v. City of Flint, 92 F.3d 396, 405 (6th Cir. 1996) (quoting O’Donnell Constr. Co. v. District of Columbia, 963 F.2d 420, 427 (D.C. Cir. 1992)). Here, Plaintiffs’ anecdotal evidence fails to raise any inference that KPMG engaged in a systemwide pattern or practice of discrimination. The record contains the 150 internal complaints by or on behalf of putative class members submitted to KPMG’s Human Resources and Ethics & Compliance departments, including 147 complaints of sexual harassment and “unprofessional behavior,” and 134 complaints of gender discrimination. Plaintiffs also submitted 20 declarations from putative class members describing misconduct at KPMG and presumably would have submitted many more had the Court not limited the parties’ submissions.14 The issue is whether KPMG paid and promoted women and men differently throughout the firm because of discriminatory intent. If Plaintiffs had presented strong statistical evidence of discriminatory treatment, then anecdotal evidence might bolster the inference that the discriminatory treatment was the result of discriminatory intent. However, in this case as discussed above, Plaintiffs lack strong statistical evidence showing that women and men were paid and promoted differently. Anecdotal evidence of individual incidents of sexist attitudes and behavior -- whether 200 or 500 in number -- cannot supply this critical missing piece. See Dukes, 564 U.S. at 358 (“In Teamsters, in addition to substantial statistical evidence of company-wide 14 In Teamsters, the plaintiffs offered 1 anecdote for every 8 class members to buttress the statistical evidence of discriminatory treatment. See Dukes, 431 U.S. at 358. Here, Plaintiffs would have needed 1,250 anecdotes for the approximately 10,000 class members to achieve the same ratio. Holding Plaintiffs to this measure, especially in cases of this size, undermines any standard of reasonableness and proportionality that should temper what is demanded of litigants, cf., e.g., Fed. R. Civ. P. 26 (“Parties may obtain discovery regarding any nonprivileged matter that is relevant to any party’s claim or defense and proportional to the needs of the case . . . .”), and undermines the Court’s need to manage the amount of evidence that will be presented for the Court’s review. See Rock v. Arkansas, 483 U.S. 44, 64 (1987) (“[A]n individual’s right to present evidence is subject always to reasonable restrictions.”). 50 discrimination, the Government (as plaintiff) produced about 40 specific accounts of racial discrimination from particular individuals.”) (citation omitted, emphasis added); see also Teamsters, 431 U.S. at 339 (“[T]his was not a case in which the Government relied on ‘statistics alone.’ The individuals who testified about their personal experiences with the company brought the cold numbers convincingly to life.”). Also, reliance on anecdotal evidence to show disparate treatment undermines Plaintiffs’ goal of showing commonality. Anecdotal evidence, by definition, raises individual rather than common questions. In Dukes, the Supreme Court found the plaintiffs’ anecdotal evidence “too weak to raise any inference that all the individual, discretionary personnel decisions are discriminatory” in part because individual managers are likely to provide non-discriminatory reasons for their respective compensation and promotion decisions. 564 U.S. at 357. In the burden-shifting framework of disparate treatment cases, the employer is entitled to elicit such evidence; the employer must rebut the presumption of discriminatory intent and show some nondiscriminatory reason for the apparently discriminatory results, and may respond with any relevant evidence. City of New York, 717 F.3d at 84–85. The prospect of Defendant rebutting the approximately 270 separate incidents Plaintiffs have introduced illustrates that the question of discriminatory intent in this case is not a common one. 3. Rule 23(c)(4) Issue Class Certification Barring certification under Rule 23(b)(2) or (b)(3), Plaintiffs seek issue class certification as to issues such as liability or disparate impact under Rule 23(c)(4). The Court declines to certify any issue class because Plaintiffs have not satisfied the commonality requirement under Rule 23(a). “[E]ven if the common questions do not predominate over the individual questions so that class certification of the entire action is warranted, Rule 23 authorizes the district court in 51 appropriate cases to isolate the common issues under Rule 23(c)(4)(A) and proceed with class treatment of these particular issues.” In re Nassau Cty. Strip Search Cases, 461 F.3d 219, 226 (2d Cir. 2006) (quoting Valentino v. Carter-Wallace, Inc., 97 F.3d 1227, 1234 (9th Cir. 1996)). But as discussed above, Plaintiffs have not identified a common question warranting class certification under Dukes, 546 U.S. at 349. Because there are no common questions that will generate “common answers” in a classwide proceeding, id. at 350, no issue class can be certified under Rule 23(c)(4). 4. New York Subclass Plaintiffs’ application to certify a New York subclass is denied because Plaintiffs fail to provide any evidence specific to a New York class. For instance, Dr. Vekker does not present any analysis of pay and promotions specific to employees in New York. His findings of nationwide aggregate disparities say nothing about New York pay or promotions. Plaintiffs argue without citing authority that they need not submit statistical analysis specific to the application of KPMG’s national employment policies to New York employees at the class certification phase. Even assuming that this argument is valid, it offers no reason why a New York class can be certified when, as discussed above, the record is insufficient to show the commonality necessary for certification of a nationwide class. C. EPA Final Certification Plaintiffs move for final certification of an EPA collective, which was conditionally certified in 2014, of “all female employees of KPMG who held any of the following positions -Associate, Senior Associate, Manager, Senior Manager/Director and/or Managing Director -- in either the Tax or Advisory practice groups from October 17, 2008, to the present” (the “Proposed Collective”). Kassman v. KPMG LLP, No. 11 Civ. 3743, 2014 WL 3298884, at *9 (S.D.N.Y. 52 July 8, 2014). Final certification of the Proposed Collective is denied because Plaintiffs fail to show that the members of the Proposed Collective (1) work at a single “establishment” and (2) are “similarly situated” to one another. 1. Background Legal Framework The EPA prohibits an employer from paying wages “at a rate less than the rate at which he pays wages to employees of the opposite sex in such establishment for equal work on jobs the performance of which requires equal skill, effort, and responsibility, and which are performed under similar working conditions . . . .” 29 U.S.C. § 206(d)(1). “As part of the [Fair Labor Standards Act (the “FLSA”)], the EPA utilizes the FLSA’s enforcement mechanisms and employs its definitional provisions.” Anderson v. State Univ. of N.Y., 169 F.3d 117, 119 (2d Cir. 1999), vacated on other grounds, 528 U.S. 1111 (2000). Section 216(b) of the FLSA provides that an EPA action “may be maintained against any employer . . . in any Federal or State court of competent jurisdiction by any one or more employees for and in behalf of himself or themselves and other employees similarly situated.” 29 U.S.C. § 216(b). “Neither the FLSA nor its implementing regulations define the term ‘similarly situated.’” Cunningham v. Elec. Data Sys. Corp., 754 F. Supp. 2d 638, 643 (S.D.N.Y. 2010) (internal quotation marks omitted). “Certification” under the FLSA is simply “the district court’s exercise of the discretionary power . . . to facilitate the sending of notice to potential class members.” Myers v. Hertz Corp., 624 F.3d 537, 555 n.10 (2d Cir. 2010). Although the Second Circuit has never offered a definitive standard for the certification of collective actions under the FLSA, it has endorsed the two-step approach widely used by the district courts in this Circuit and by other circuit courts. See id. at 554–55 (“[T]he district courts of this Circuit appear to have coalesced around a twostep method, a method which . . . we think is sensible.”). 53 “The first step involves the court making an initial determination to send notice to potential opt-in plaintiffs who may be ‘similarly situated’ to the named plaintiffs with respect to whether a FLSA violation has occurred.” Id. at 555. To establish that the named plaintiffs are “similarly situated” to the potential opt-in plaintiffs in the first stage of the inquiry, the named plaintiffs must “make a ‘modest factual showing’ that they and potential opt-in plaintiffs ‘together were victims of a common policy or plan that violated the law.’” Id. (quoting Hoffmann v. Sbarro, Inc., 982 F. Supp. 249, 261 (S.D.N.Y. 1997)). “Once a court is satisfied that the first-stage inquiry has been satisfied, it conditionally certifies the class and orders notice to putative class members, who are given the opportunity to opt in.” Cunningham, 754 F. Supp. 2d at 644. The second step, typically taken upon the completion of discovery, requires the court to determine, “on a fuller record, . . . whether a socalled ‘collective action’ may go forward by determining whether the plaintiffs who have opted in are in fact ‘similarly situated’ to the named plaintiffs. The action may be ‘de-certified’ if the record reveals that they are not . . . .” Myers, 624 F.3d at 555. 2. Single Establishment The EPA requires equal pay for equal work for employees who work in a single “establishment.” 29 U.S.C. § 206(d)(1). The term “establishment” in the EPA “refers to a distinct physical place of business rather than to an entire business or ‘enterprise’ which may include several separate places of business.” 29 C.F.R. § 1620.9(a). However, “unusual circumstances may call for two or more distinct physical portions of a business enterprise being treated as a single establishment.” 29 C.F.R. § 1620.9(b). “For example, a central administrative unit may hire all employees, set wages, and assign the location of employment; employees may 54 frequently interchange work locations; and daily duties may be virtually identical and performed under similar working conditions.” Id. Courts identify “a widely followed standard recognizing that central control and administration of disparate job sites can support a finding of a single establishment for purposes of the EPA.” Mulhall v. Advance Sec., Inc., 19 F.3d 586, 591 n.11 (11th Cir. 1994) (collecting cases). “The hallmarks of this standard are centralized control of job descriptions, salary administration, and job assignments or functions.” Id. at 591–92 (holding that “[a] reasonable trier of fact could infer that because of centralized control . . . a single establishment exist[ed] for purposes of the EPA”). Courts reason that “narrow construction of the word ‘establishment’ could make proof of discrimination more difficult, thus frustrating congressional intent.” Id. at 591 (internal quotation marks omitted). “However, [courts] presume that multiple offices are not a ‘single establishment’ unless unusual circumstances are demonstrated.” Meeks v. Comput. Assocs. Int’l, 15 F.3d 1013, 1017 (11th Cir. 1994) (citing 29 C.F.R. § 1620.9(a)). In Meeks, the Eleventh Circuit ruled that “the evidence . . . [did] not demonstrate the level of centralization necessary to justify treating all of the company’s technical writers as working at a single establishment” where “[a]lthough [defendant] centrally sets broad salary ranges, the specific salary to be offered a job applicant is determined by the local supervisor,” and “[a]lthough personnel records are maintained centrally, job applicants are interviewed by local officials and hired upon their recommendation.” Id. As discussed above, upon review of a “fuller record,” Myers, 624 F.3d at 555, although KPMG set generally applicable guidelines, individual pay and promotion decisions were left to the discretion of local practice area leaders. These decentralized discretionary determinations were reviewed by firm leadership on an aggregate basis against budget. Pay and promotion 55 decisions were not sufficiently “centralized” to amount to “unusual circumstances” warranting a finding that the many offices and practice areas represented in the 1,100-member Proposed Collective qualify as a single “establishment” under the EPA. See Meeks, 15 F.3d at 1017; 29 C.F.R. § 1620.9(b). 3. “Similarly Situated” At the final certification stage, “the district court will, on a fuller record, determine whether a so-called ‘collective action’ may go forward by determining whether the plaintiffs who have opted in are in fact ‘similarly situated’ to the named plaintiffs.” Myers, 624 F.3d at 555. In analyzing this issue: Although the Second Circuit has yet to prescribe a particular method for determining whether members of a class are similarly situated, district courts in this circuit typically look to the (1) disparate factual and employment settings of the individual plaintiffs; (2) defenses available to defendants which appear to be individual to each plaintiff; and (3) fairness and procedural considerations counseling for or against collective action treatment. Zivali v. AT & T Mobility, LLC, 784 F. Supp. 2d 456, 460 (S.D.N.Y. 2011) (alterations and internal quotation marks omitted); accord LoCurto v. AT&T Mobility Servs. LLC, No. 13 Civ. 4303, 2018 WL 4519201, at *3 (S.D.N.Y. Sept. 20, 2018). All three considerations weigh against certifying the Proposed Collective. First, Plaintiffs seek to certify a Proposed Collective with approximately 1,100 members, from approximately 80 offices nationwide, who work in about 115 different Cost Centers. As illustrated by Dr. Banks’s report, the job content of the 1,100 Opt-Ins is wide-ranging. The sheer size and breadth of the Proposed Collective suggests that its members are not similarly situated. See, e.g., Scott v. Chipotle Mexican Grill, Inc., No. 12 Civ. 8333, 2017 WL 1287512, at *8 (S.D.N.Y. Mar. 29, 2017) (“[T]he ‘similarly situated’ analysis for the purposes of FLSA 56 certification can be viewed, in some respects, as a sliding scale. In other words, the more opt-ins there are in the class, the more the analysis under § 216(b) will mirror the analysis under Rule 23.”). The second and third factors lead to the same conclusion. For the reasons discussed above with respect to commonality, there is no (non-discretionary) uniform causal mechanism for determining pay and promotion operating across the Proposed Collective. This means that there are likely 1,100 defenses to justify why the 1,100 Opt-Ins were paid as they were. Adjudicating the claims of the proposed collective in a single action would give rise to obvious procedural difficulties and could not assure fair treatment of any party involved. Despite the size of the Proposed Collective, and variety of job responsibilities amongst Opt-Ins, Plaintiffs nevertheless argue, “Because Opt-In Plaintiffs within the same job level and function perform the same principal duties under similar working conditions, certification of this collective is appropriate. Any marginal variation between Opt-in Plaintiffs’ job duties does not undermine the collective nature of the claims.” In support of this argument, Plaintiffs state that “[a]ll Tax employees are expected to be able to distinguish tax authorities, identify and analyze tax issues, and use tax technologies.” Plaintiffs state, “Employees in Advisory also perform work with a common set of skills and responsibilities. They help clients improve business performance, leverage information technology, protect assets, and manage risk.” Plaintiffs also note that more senior employees across both Tax and Advisory are expected to “oversee the work of more junior staff” and “provid[e] feedback.” This argument fails, because it is shortsighted; Plaintiffs ignore what the members of the Proposed Collective will eventually need to prove -- and the impossibility of practically adjudicating that proof in a single action. The EPA requires “equal pay for -- emphatically -equal work. To that end, Congress rejected statutory language encompassing ‘comparable work’ 57 . . . .” E.E.O.C. v. Port Auth. of N.Y. & N.J., 768 F.3d 247, 254 (2d Cir. 2014) (citing 29 U.S.C. § 206(d)(1)). “[T]o prove a violation of the EPA, a plaintiff must demonstrate that ‘(1) the employer pays different wages to employees of the opposite sex; (2) the employees perform equal work on jobs requiring equal skill, effort, and responsibility; and (3) the jobs are performed under similar working conditions.’” Id. at 254–55 (alterations and internal quotation marks omitted). Assessing the second prong requires a “focus on job content,” in which courts assess “the congruity and equality of actual job content between the plaintiff and comparator . . . .” Id. at 255. “To satisfy this standard, a plaintiff must establish that the jobs compared entail common duties or content, and do not simply overlap in titles or classifications.” Id.; see also 29 C.F.R. § 1620.13(e) (“Application of the equal pay standard is not dependent on job classifications or titles but depends rather on actual job requirements and performance.”). Second Circuit “cases as well as the EEOC’s regulations and Compliance Manual stand for a common principle: a successful EPA claim depends on the comparison of actual job content; broad generalizations drawn from job titles, classifications, or divisions, and conclusory assertions of sex discrimination, cannot suffice.” Port Auth., 768 F.3d at 256 (citing EEOC Compliance Manual § 10-IV(E)(2) (2000)). Plaintiffs’ arguments fail because they operate at the wrong level of abstraction; the “equal work” analysis requires a micro-level -- rather than a macro-level -- approach to comparing job responsibilities, skills and requirements. See id. at 257 (“[B]land abstractions -untethered from allegations regarding Port Authority attorneys’ actual job duties -- say nothing about whether the attorneys were required to perform ‘substantially equal’ work.”). That female supervisors across Tax and Advisory are expected to “provide feedback” does not mean that they are similarly situated to male supervisors who also must “provide feedback.” The “equal work” analysis must “focus on job content.” Id. at 255 (emphasis added). By lumping all Tax 58 Associates into the same category, the Proposed Collective assumes without reason that an Associate in the Accounting Methods and Credits practice, who has a science Ph.D. and collects data at oil facilities in order to analyze Research and Development tax credits, performs “equal work” to an Associate in the Tax Controversy Services practice, who has a J.D., and advocates in IRS disputes. While both associates might need to “analyze tax issues,” they do not engage in “equal work” when “job content” is examined. Plaintiffs’ attempt to shoehorn the Opt-Ins into a single EPA collective asks too much when compared with other proposed collectives that courts in the Second Circuit have rejected. See, e.g., id. at 257 (holding that the EEOC failed to state an EPA claim where it pleaded only that female non-supervisory attorneys at the Port Authority were paid less than male nonsupervisory attorneys, because it cannot be assumed that “an attorney is an attorney is an attorney”); Chiaramonte v. Animal Med. Ctr., 677 F. App’x 689, 691–92 (2d Cir. 2017) (summary order) (holding that a female primary care veterinarian did not establish that she performed “equal work” to male specialist veterinarians); Miller v. City of New York, No. 15 Civ. 7563, 2018 WL 2059841, at *4 (S.D.N.Y. May 1, 2018) (“The stark differences in training, job requirements, and job responsibilities between Traffic Enforcement Agents and School Crossing Guards warrant summary judgment on the EPA claim.”); Scott v. Chipotle, 2017 WL 1287512, at *8–9 (decertifying a collective of “Apprentices” at Chipotle restaurants because “Apprentices had vastly different levels and amounts of authority in exercising managerial tasks” across the 37 states in which they worked). In short, because employees sharing a single job title at KPMG are not “similarly situated” to one another, if the Proposed Collective were certified, it would be necessary to: (1) match the Opt-ins with 1,100 unique proposed male comparators; (2) adjudicate approximately 1,100 “equal work” questions and only then (3) turn to any individualized 59 defenses KPMG might raise with respect to specific Opt-Ins (e.g., poor performance reviews). Such an undertaking is impractical if not impossible. “[F]airness and procedural considerations counsel” against collective action treatment in this case. See Zivali, 784 F. Supp. 2d at 464. For this reason too, the motion for final certification of a collective is denied, and the collective that was preliminarily certified is decertified. 60 III. CONCLUSION For the foregoing reasons, Plaintiffs’ motions for Rule 23 class certification and EPA collective certification are DENIED. Both parties’ Daubert motions and motions to strike are DENIED. The Clerk of Court is directed to close the motions at Dkt. Nos. 797, 800, 802, 803, 813, 815, 817, 818, 827 and 829. Dated: November 30, 2018 New York, New York 61

Some case metadata and case summaries were written with the help of AI, which can produce inaccuracies. You should read the full case before relying on it for legal research purposes.

This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.