Wednesday, July 06, 2011
Wal-Mart v. Dukes and the Heterogeneity Problem: Part I (the Aggregability/Resolvability Distinction)
Thanks to Dan for inviting me to be a July guest. I will first discuss the Supreme Court's recent decision in Wal-Mart v. Dukes reversing certification of a nationwide class alleging that Wal-Mart violated Title VII by discriminating on the basis of sex in pay and promotions. The decision has already received ample and insightful academic scrutiny, including from Sergio Campos on this blog. I will try to avoid redundancy by focusing on aspects of what I call the 'Heterogeneity Problem' in collective adjudication. In particular, this post highlights two distinct strands of the problem—Aggregability and Resolvability—that courts and commentators often blur, and that are helpful in assessing Wal-Mart's implications. I will also explore a subtle dimension of the Court's analysis of "dissimilarity" and class cohesion that has largely escaped notice, and that may influence how courts apply Rule 23 in future cases. The scope of Rule 23 in turn affects the judiciary's role in enforcing a wide range of private entitlements in fields such as securities law, civil rights, mass torts, and consumer protection.
A central problem in all proposed or certified class actions involves identifying and coping with heterogeneity among the class. At a high level of abstraction, classes often seem materially homogenous. The carefully worded class definition might appear to encompass a group in which each member suffered an identical injury caused by the same actor in the same manner that creates equivalent entitlements to a remedy. If scrutiny confirms that homogeneity is present, then class certification will generally be appropriate: the substantial economy of scale that collective adjudication of identical claims provides often justifies the cost and complications of group litigation.
The Heterogeneity Problem arises because substantive legal rules frequently require detailed inquiries that reduce the operational level of abstraction and expose differences between ostensibly homogenous class members.
For example, individual people who meet the class definition might have behaved differently, encountered different types of conduct by the defendant, suffered different kinds or degrees of injury, or possessed different levels of justification for their conduct in light of their knowledge and circumstances. Alternatively, the class may define similarly situated people, but identifying those people might require individualized inquiries. Consider a hypothetical class action against a pharmaceutical manufacturer alleging that drug X causes heart attacks in a subset of users. Assume that every victim of X-induced heart attacks has a valid claim under the applicable law, that victims of heart attacks caused by other factors do not have valid claims, that compelling evidence establishes who ingested the tightly controlled drug and which of them suffered heart attacks, but that evidence of causation is debatable in each case. Now imagine two class definitions: (1) "all people who ingested Drug X" and (2) "all people who ingested Drug X and as a result suffered a heart attack." The first definition encompasses people without valid claims, while everyone in the second class has a valid claim, yet is not easily identifiable. Both definitions encounter essentially the same heterogeneity problem: distinguishing people with valid claims from those without valid claims requires an individualized inquiry. Playing shell games with the class definition shuffles the heterogeneity problem around, but cannot make it disappear.
Coping with heterogeneity in class actions can be costly. Heterogeneity may require holding numerous hearings, modifying otherwise applicable procedures to streamline litigation, prioritizing the interests of some class members at the expense of others, challenging the trier-of-fact's cognitive ability to distinguish dissimilar claims presented concurrently, or tolerating settlements that emerge more from the defendant's fear of a large judgment than acknowledgment of actual liability to all class members. At some point, heterogeneity becomes impossible to manage in a single proceeding without either undue burden or sacrificing values that the judicial system deems sufficiently central to the enterprise of public adjudication. Identifying that tipping point by defining the relevant values and determining their elasticity implicates interrelated concerns at the heart of Rule 23. When heterogeneity is manageable, class actions can be costly but useful procedural tools for enforcing substantive rights; when heterogeneity is excessive, class actions are not viable.
Deciding when heterogeneity is excessive requires two distinct inquiries: determining when heterogeneity exists and determining how heterogeneity matters. The first inquiry requires determining what class members allege and how they must prove their claims in light of the applicable substantive law. The goal of this inquiry is to see how much the class really has in common. This inquiry overlaps with assessment of claims on the merits, merging the pleading, certification, and summary judgment stages of litigation in awkward ways that Richard Nagareda addressed in an article on "Aggregate Proof" (which the Court cited in Wal-Mart). See Richard A. Nagareda, Class Certification in the Age of Aggregate Proof, 84 N.Y.U. L. Rev. 97 (2009). I call this the Aggregability inquiry, as it requires considering when a court may provide the same answer to multiple iterations of similar questions raised by different people. The court must ask, in essence, whether issues or claims are aggregable based on the record and the governing law. The second inquiry requires determining if courts can accommodate non-aggregable components of a case within a class action setting. In prior work I have called this the "Resolvability" inquiry because it requires assessing whether individualized questions preclude final resolution of the entire case. See Allan Erbsen, From "Predominance" to "Resolvability," 58 Vand. L. Rev. 995 (2005).
The distinction between Aggregability and Resolvability is useful in analyzing Wal-Mart. The Court in Wal-Mart primarily addressed Aggregability, but may have laid the groundwork for significant changes in how lower courts address Resolvability.
In Wal-Mart, the Heterogeneity Problem arose because a large number of employees requested an injunction against allegedly discriminatory employment practices. The plaintiffs sought to find commonality in the allegation that they were all victims of a tainted corporate culture that infected ostensibly decentralized decisionmaking. But the Court disagreed, concluding that the plaintiffs had not identified a "general policy of discrimination" and therefore had not established even a single classwide common question that could justify certification under Rule 23(a)(2)'s "commonality" requirement.
The Court's holding primarily addresses Aggregability. The majority concluded that the record revealed distinctions that the substantive law deemed meaningful. This heterogeneity precluded the district court from providing identical answers to questions raised by different class members. Commentators have extensively discussed this holding, often faulting the Court for not reading Title VII in a way that would vindicate its purposes. Their argument generally posits that Title VII aspires to eradicate discrimination, that discrimination often manifests in subtly different ways as corporate culture permeates a decentralized corporate structure, and that effective statutory enforcement therefore requires providing remedies that look past variations in individual facts to the basic underlying problem. Criticism of the Court's approach to rights and remedies cuts to the heart of the Aggregability problem, which entails deciding when substantive law should adapt to the procedural realities of the environment in which it operates. The Wal-Mart holding is thus important for what it says about Title VII, and for what it implies about the Court's potentially stingy approach to Aggregability in other contexts.
Wal-Mart's Aggregability holding is controversial, but its Resolvability holding is not. Accepting the Aggregability holding (for the sake of argument) requires concluding that the case was not resolvable as a class action. If the plaintiffs did not raise even one common question, then collective adjudication would serve little purpose in light of its costs. The Resolvability problem is interesting only when some questions are common to the class, and some are not. The common questions weigh in favor of certification, while the individualized questions weigh against certification. At that point, courts must decide whether Rule 23 permits bundling the homogenous and heterogeneous aspects of the case into a single class proceeding (or cluster of proceedings). Wal-Mart was thus an outlier case—a rare example of a purely heterogeneous class that nevertheless obtained certification (at least according to the Court's assessment of the record and applicable law, which I am not analyzing here). The holding in Wal-Mart is thus not really "about" class actions as a procedural device so much as it is about how substantive law accommodates the available procedural methods for enforcing it.
However, Wal-Mart may say more about Resolvability than is apparent at first glance and that may generate future controversy. In several instances, the Court referenced the concept of "dissimilarity" between class members as an obstacle to collective adjudication. I will consider the significance of this language in a subsequent post.
TrackBack URL for this entry:
Listed below are links to weblogs that reference Wal-Mart v. Dukes and the Heterogeneity Problem: Part I (the Aggregability/Resolvability Distinction):
I really enjoyed your post and appreciate the shout out. What if, like me, you view the class action as a trust device. That it, it allows the plaintiffs to invest in common issues in the same way as the defendant. If the class action has this function, then isn't the mere existence of common issues (along with numerosity) sufficient for class certification? Moreover, couldn't the heterogeneous aspects of the litigation be treated through bifurcation or sampling, which courts used for Title VII cases prior to Wal-Mart? Again, great post, and great article too!
Posted by: Sergio Campos | Jul 6, 2011 1:21:57 PM
Sergio, thanks for your comments and prior posts on Wal-Mart. On your first point, I agree that class actions help overcome collective action problems and enable a class to match the defendant’s incentives to invest in a case (although the stakes usually will not be fully symmetric because the investors in the plaintiffs’ claims—generally class counsel—would receive only a percentage of the recovery, while the defendants would be on the hook for the full judgment). However, the fact that class actions overcome collective action problems explains in part why they can be useful, but not how they can appropriately be deployed. The heterogeneity problem gets at this latter question of when class adjudication is feasible and consistent with values animating civil adjudication. On your second point, bifurcation and sampling can be useful and effective tools, but with a caveat: if the defendant is unwilling to settle claims based on the results of sample trials or common issue hearings, then those results can translate into a judgment only with further individualized proceedings, which might not be feasible (raising a Resolvability problem), or by court order, which the substantive law might not allow (raising an Aggregability problem). So I don’t mean to suggest that any dollop of heterogeneity is always fatal to certification. As you indicate, judges can creatively manage even complex cases. Nevertheless, heterogeneity presents problems that require thinking carefully about both Resolvability and Aggregability before making a certification decision.
Posted by: Allan Erbsen | Jul 6, 2011 11:09:18 PM
Bifurcation and sampling ahave many problems, in my view. The Court was unanimous in rejecting the sampling in Wal-Mart. So that's something that can't be laid at the feet of the Scalia 5; it's the Ginsburg 4 as well. That rejection raises serious questions about whether the seeming chasm between the 5 and 4 is all that great. Ginsburg would have allowed the class to try again under 23(b)(3) as opposed to 23(b)(2). But would that have gotten far, given the problems that arise under 23(b)(3) and damages claims?
How would you bifurcate and handle the jury? Have the same jury sit for 3 years to hear every mini-trial? I see no authority for having a "class jury" resolve the common issue, but then having separate juries hear each employee-plaintiff's claim. Further, it's not just a matter of separating damages from liability, as even the liability finding is ultimately individualized as a matter of substantive Title VII law. Even if the company discriminated in general, Jane Doe #45 has to show that SHE suffered from that discrimination (given the rejection of sampling, per all 9). What if she never applied for promotion, or wasn't qualified, etc.
That makes it seem to me that Ginsburg's remand option would end up a dead end, given the parts she (and all 4) joined. In my view, much of the commentary has been deficient in not addressing this. Your discussion of Resolvability and Aggregability hits these issues, but I'd be curious about how you the decision itself in light of your analysis. Do you believe that the Ginsburg view, had it been the majority, could have really led to a viable class, or would it have merely pushed back the day of reckoning in this case (and made a diffrence for other cases)?
Posted by: class action geek | Jul 8, 2011 3:10:20 PM
class action geek: I agree that bifurcation can sometimes, but not always, merely postpone rather than avoid what you call “the day of reckoning.” On the one hand, a defendant who loses a classwide liability hearing might have strong incentives to settle individual claims rather than prolong public scrutiny of its conduct. These settlements would resolve the litigation and vindicate the certification order (assuming that the liability finding was correct on the merits and that the settlements are fair). On the other hand, if the defendant refuses to settle and insists on fighting the remaining aspects of each individual claim, and if the applicable substantive law permits this tactic, then the plaintiffs might be unable to translate their initial victory into a tangible reward, and certification may in hindsight appear inefficient. The difficult question is whether at the time of certification judges should be able to use the uncertain prospect of a settlement as a justification for bifurcation in circumstances where individualized proceedings would not be feasible. I discuss this problem in more detail at pages 1030-33 and 1082 of my Resolvability article.
Posted by: Allan Erbsen | Jul 8, 2011 5:29:33 PM