Some Important Class Action Guidance for Discrimination Cases

As previously predicted (you read it here first!!), plaintiffs’ lawyers are moving from litigating individual plaintiff discrimination cases to attempting to craft wide-ranging class actions based on disparate treatment claims. Despite the Supreme Court’s decision in the Dukes v. Walmart case last summer, we are seeing more and more attempts to link hundreds of plaintiffs together in huge cases that will be expensive nightmares to defend.

Of course, that’s the point.  Most employers, when confronted by a lawsuit that will cost them more than a million dollars just to prepare for trial, will immediately settle. That’s why it is so important to understand what the courts consider to be valid classes, and think about how a company’s employment practices can be structured and make it as difficult as possible to create such a class.

A recent federal case (note:  Pacer needed for viewing) out of Illinois provides some useful guidance in the creation of disparate treatment and class actions. Twelve construction workers originally filed a case against a large construction company alleging race discrimination across the company. In a shotgun approach, the Plaintiff sought to certify four separate classes – a hostile work environment class, (which encompassed all black employees over a period of approximately 10 years), a hiring and promotion class (comprising all black employees denied hiring , rehiring, or promotions), a work hours and compensation class (made up of black employees denied opportunities to work, not afforded overtime hours, or premium pay hours), and a layoff and termination class (comprising all black employees laid off or terminated). The potential class size ranged in the thousands of employees, and because of the extended duration of the case, presumed to be a litigation hornets nest for the company.

There are several variations of class actions at play in this case, all under Rule 23 of the Federal Rules of Civil Procedure. Class actions have two key requirements – numerosity and commonality. Specifically, there must be so many potential plaintiffs affected by a common practice or procedure that adjudication of all the claims together is more efficient and proper than try each claim separately. In disparate treatment discrimination cases, class certification traditionally has been very difficult because each employment decision tends to be the result of highly specific factual considerations by an individual manager. Disparate impact class actions are far easier to fit into the class action scheme because a plaintiff doesn't have to prove discriminatory intent – the only inquiry the court makes is whether there is a legitimate basis for the employment policy that has an unintentional impact on a protected class.

With respect to the question of commonality – i.e. is there a question of law or fact that is common to all of the black construction workers – the court looked at the company’s practice of delegating to individual jobsite superintendents discretionary authority to make employment decisions. In assessing the propriety of the class claim, the trial court determined that distinguishing characteristics of the Walmart decision was that the size of the potential class (over 1 million employees) overwhelmed any potential common issue based on delegation of authority. Because the numbers in this case were not nearly so significant, the court determined that there could be a class affected by the company’s policy of delegating discretionary authority to job site superintendents, if the plaintiffs can show evidentiary basis  for their class definition.

I don’t think the court did a good job of differentiating Walmart from this case, but the issue was not significant because the court determined that the plaintiff’s statistical evidence discrimination was insufficient to certify the class. Although the plaintiffs pointed to some evidence that there was a company wide difference in the distribution of jobs and overtime for African Americans, it could not refine that evidence down  to show there was a common practice of intentional discrimination at the individual work sites. In fact there was only limited anecdotal evidence of discriminatory work hours and overtime practices at 7 of the 262 construction sites that the company operated.  Discriminatory layoff and termination practices contained anecdotes from only 5 construction sites. This was not enough to show that there was a common question as to whether the delegation of authority to supervisors at all of the company’s sites resulted in discrimination against African American employees and so the court did not certify the disparate treatment class.

As alluded to above, the disparate impact class, which included a claim focusing on assignment of work hours, overtime, and terminations, faired better in the court’s analysis. Because the plaintiffs did not have to show that there was an intent to discriminate, the wide discretion allowed to site managers appeared to result in significantly lower overtime and work hours to black employees and created an issue for class resolution.  The court certified a class with respect to the work hours claims, but gave no credence to the statistical analysis relating to hiring and termination data. The court’s discussion of the shortcomings of the Plaintiffs’ statistical expert are noteworthy because they highlighted the fact that the Plaintiffs’ expert failed to properly assess the availability of qualified black construction workers in the target metropolitan area. The court removed that evidence from its consideration and the resulting limited anecdotal evidence of hiring and termination was not sufficient to show a class wide pattern.

The court certified a class subject to hostile work environment because it found that the record contained “voluminous” evidence alleging that supervisors at two work sites engaged in overtly racist conduct that was likely known to company management . This fact alone, even though it was only identified at two sites, was enough for the court to find that there was sufficient evidence to certify a common issue of fact for all black construction workers throughout the company.

The court next determined that the claims of the identified class representatives were sufficient to allow them to proceed under the disparate impact and hostile work environment claims although the court determined that there were no remaining plaintiffs that were denied a promotion so that this claim was invalid. The court then ruled that the parties could not certify a Rule 23(b)(2) class (which is typically an easier class to certify) because there was no overriding claim for injunctive relief. Instead, the court determined that money damages were the major remedy sought by the class, and held the employees to a stricter Rule 23(b)(3) class standard. Under this analysis, the court determined that the disparate impact claims with respect to work hours and compensation were proper, as were the hostile work environment claims. Those claims would be allowed to go forward  as class claims through discovery and trial.

What are the lessons here?

The first is that, for better or worse, the courts are not enamored of so called “push down” management styles in which a company provides general guidance and unlimited discretion to low level managers to run their own operations. Systematic and regular reviews of hiring practices, promotion practices, and terminations and lay offs are absolutely essential for large companies that provide little in the way of continuing oversight of their remote operations. Second, companies absolutely can not turn a blind eye to potential discrimination issues in their supervisory work force. The main reason why the hostile work environment claim was certified here, even though it was only identified at 2 work sites, is that there was evidence that management was aware of what was going on. Management’s tolerance of blatant racist conduct (or at least alleged racist conduct) at two sites meant that the court could certify the hostile work environment across all the work sites, because of the presumption  such conduct wasn't an issue for the company management.

In other words, the Dukes v. Walmart legacy is going to be limited to cases where there are huge numbers of employees that make a system wide adjudication almost impossible. In all other circumstances, companies need to look carefully at what’s coming out of the employment policy pipeline when they are giving large amounts of discretion to local management officials.

NLRB Limits the Scope of Arbitration Agreements

In what is being perceived as a part of a continuing attack on the ability of employers to limit their litigation exposure through employment agreements requiring arbitration (and also part of a continuing effort to bolster the business of labor unions and trial lawyers), the NLRB has now determined that employers may not require their employees to sign arbitration agreements limiting class actions or other collective legal claims. The decision, if it goes unchallenged on appeal, will have significant implications for employers that use arbitration agreements as a means of limiting their class-action exposure.

The Board's decision appears to fly in the face of a recent Supreme Court case which held that state courts could not prohibit class-action waivers in arbitration agreements. That case--AT&T Mobility--determined that limiting the scope of an agreement in a consumer contract frustrated the Federal Arbitration Act's purpose of encouraging arbitration of disputes.

The agreement in question here required all disputes and claims relating to employment issues to be submitted to final and binding arbitration, and the arbitrator was limited to hearing only individual claims. The Board determined that the effect of the agreement was to limit employees, as a condition of employment, from filing collective litigation of claims in any form, arbitral or judicial.

The Board determined that the agreement's provision violated the National Labor Relations Act's defense of protected concerted activity by employees; so-called Section 7 rights. Section 7 rights are particularly broad, and apply regardless of whether the employer is unionized. The Board has greatly expanded the scope of Section 7 rights recently, particularly in the arena of social media. Accordingly, all employers that make use arbitration agreements need to pay careful attention to this holding.

In an interesting rhetorical twist, the Board noted that employees represented by a union can waive their collective action rights under the terms of a collective-bargaining agreement, but that individual employees may not. The Board also noted that AT&T Mobility applied to consumer contracts; this was a distinction the Board found important because the Board's decision affected far fewer people in each instance than a typical consumer contract.

This distinction seems to me to be a false one-the scope of the Board's efforts arguably apply to more than 120 million people, far more than an average consumer contract. Moreover, there is Supreme Court precedent indicating that there is no particular difference between arbitration agreements signed by individual employees and those signed by union representatives.

I should also note here that the Board's decision does not foreclose the use of arbitration agreements per se. As long as an employer leaves open a way for its employees to engage in class action litigation, the employer may require individual claims to be arbitrated. Of course, this option keeps in place the ability of employees and their attorneys to engage in hugely expensive collective actions, about which the employer can do nothing.

The employer in this case has the ability to appeal this decision, and I suspect it will do so. In the meantime, one possible way of limiting collective action exposure would be to write a detailed arbitration agreement limiting discovery and litigation processes, which are the most significant expenses in class action litigation. Whether that would survive Board review is an open question.  Finally, employers should note that managerial employees are not covered by the Board's decision, nor are independent contractors, an important exception for many employers.

If at First You Don't Succeed…

Sue, sue, again. At least that's the approach of the plaintiffs' litigation team in the late and unlamented Dukes v. Walmart litigation that was unceremoniously bounced from the ranks of class-action cases by the Supreme Court last year. You may recall that the Court determined that the Ninth Circuit's approval of the class of approximately 1.5 million women who worked at Walmart during the relevant period was inappropriate and improvident. The majority on the Court focused on the allegation in the class certification that the plaintiffs were all similarly affected by Walmart's centralized policy of decentralization that allowed individual store managers to make employment decisions based on a scheme affected by centralized and pervasive anti-woman bias.
If that sounds like unmitigated lawyer doubletalk, then you agree with Justice Scalia and the rest of the majority.
The plaintiffs' law firm has now refiled the case, this time on behalf of only 90,000 current and former female employees who work for Walmart in California. But this doesn't seem to solve the problem mentioned above-that if these decisions were decentralized, it's almost per se impossible to certify a class based on the resulting treatment. In fact, this will be Walmart's defense in this case--namely that each individual employment decision, or at least each individual store manager's employment decisions, will stand on their own and cannot provide the basis for such a wide-ranging class.
The plaintiffs are alleging that they have new statistical evidence that was not put before the Supreme Court in the original litigation. Short of some kind of clear link between these thousands of employment decisions at issue, plaintiffs may find it's "class dismissed", even in the relatively employee-hospitable environs of the Ninth Circuit.

A Tale of Two Class Actions

The recent decision by Judge Loretta Preska in New York that blew up the EEOC’s discrimination case against Bloomberg Financial News Service is worth reading, despite its length, for what it says about large scale employment discrimination cases. It's particularly worth reading along with the Supreme Court’s decision in Dukes v. Walmart, another large scale case that fell apart after a rational judicial assessment of the basis for the class claims.  

Both cases involved large numbers of female class members alleging that gender discrimination was the norm at their employers. The Walmart group was gigantic – more than a million members – and was brought as a true class action. Bloomberg, on the other hand, involved only some 600 employees who took maternity leave during a 7-year time period (the EEOC claimed Bloomberg discriminated against pregnant women and mothers). Both cases, however, alleged that individual employment decisions made by Walmart and Bloomberg management were permeated with bias--against women generally, or against pregnant women in particular. Both cases ultimately fell apart because of the inability of the plaintiffs to provide valid statistical evidence to support their claims and because the plaintiffs relied on  individual anecdotal statements that ultimately could not support the wide scale relief that they sought.

The plaintiffs  used different approaches in each case. As noted above, the Walmart plaintiffs filed their case under traditional class action principles; the EEOC went after Bloomberg using a quasi- class action process by alleging a “pattern or practice” charge, and claiming that it was the normal process at Bloomberg to discriminate against pregnant employees.

Both plaintiffs groups were betting that they could overcome the normal problems associated with  filing discrimination claims on behalf of large numbers of employees by using a mix of unreliable statistical comparators, coupled with what were supposed to be fairly graphic claims of employment discrimination against individual class members. Although the Walmart plaintiffs were successful in getting passed the 9^th Circuit with their statistical nonanalysis, the Supreme Court  made short work of it on review. The Court noted that for the statistics to have any meaning, the individual plaintiffs must have been subjected to a  common decision-making system. In this case, what was alleged was actually a non-systematic approach to employment decisions, namely that the company delegated wide latitude to individual store managers with respect to promotions, hiring, and compensation. The Walmart plaintiffs then made the ironic logical jump (for a discrimination claim, at least) that because the decision makers were for the most part male, they would necessarily be biased against women. The Supreme Court vitiated this analysis on several levels, noting in particular that a policy of not having a policy was not, in fact, a policy. It also blew up the statistical analysis, which could not do anything more than state that it was a possibility that discrimination would result in a system with wide latitude at the individual store manager level.

In Bloomberg, the judge precluded the EEOC from even getting its statistical analysis into evidence. Amazingly, the court called the EEOC on a fundamental error in its statistical work; namely that the EEOC’s statistical expert compared women who took maternity leave not against other employees who took similar long term leaves of absence for reasons other than maternity, but rather against the work force that did not take leave at all. This is a fundamental and surprising error for a federal employment agency to make. In fact, the judge noted that it was not illegal for a company to discriminate against people who took absences for extended periods of time in terms of compensation and promotion as long as the company was not singling out people who missed work for protected reasons such as pregnancy. As the court noted, the law does not require that pregnant women be treated better than their co-workers, or given more advantages, it only requires that they not be treated worse than other employees who are not pregnant but similarly situated in their ability or inability to work. Indeed, the Civil Rights Act says as much in its section on pregnancy discrimination.

Finally, Judge Preska’s opinion should be read for what it says about the legal requirements of the modern workplace and so-called “work life balance.”  The judge cites former GE CEO Jack Welch: “There is no such thing as work life balance. There are work life choices, and you make them, and they have consequences.” The judge then went on to say that the law does not mandate that employees offer a work life balance and does not require companies to ignore an employee’s family-work  tradeoffs when deciding about employee pay and promotion. Employees who make decisions that preference family over work must understand that those decisions come with consequences, and as long as those consequences occur for anyone who takes significant time away from the company, then the company’s policies are legal.

It's unusual to get such solid and thoughtful guidance in opinions written by often harried federal judges. Judge Preska has done everyone a favor by taking the time to do so here.