George McREYNOLDS, et al., on behalf of themselves and all others similarly situated, Plaintiffs-Appellants, v. MERRILL LYNCH, PIERCE, FENNER & SMITH, INC., Defendant-Appellee, 672 F.3d 482

POSNER, Circuit Judge. The plaintiffs have filed a class action suit that charges Merrill Lynch with racial discrimination in employment in violation of Title VII of the Civil Rights Act of 1964 and 42 U.S.C. § 1981. The plaintiffs ask that a class be certified for two purposes: deciding a common issue, Fed.R.Civ.P. 23(c)(4) — whether the defendant has engaged and is engaging in practices that have a disparate impact (that is, a discriminatory effect, though it need not be intentional) on the members of the class, in violation of federal antidiscrimination law — and providing injunctive relief. Fed.

R.Civ.P. 23(b)(2). They also want damages. But while they asked the district court to certify the class for purposes of seeking compensatory and punitive damages, see Rule 23(b)(3), at argument the plaintiffs’ lawyer said she wasn’t asking— not yet anyway — for such certification, though her opening brief had suggested that if we found that the district court had erred in refusing to certify for class treatment the disparate impact issue and injunctive relief, we should order the court to “consider [on remand] the extent to which damages issues also could benefit from class treatment, consistent with Allen v.

International Truck & Engine Corp., 358 F.3d 469 (7th Cir.2004).” We defer that question to the end of our opinion. But we note here that without proof of intentional discrimination, which is not an element of a disparate impact claim, the plaintiffs cannot obtain damages, whether compensatory or punitive, but only equitable relief (which might however include backpay, and thus have a monetary dimension). 42 U.S.C. § 1981a(a)(1); Kolstad v. American Dental Association, 527 U.S. 526, 534, 119 S.Ct. 2118, 144 L.Ed.2d 494 (1999).

Section 1981a(a)(

l) is explicit that damages cannot be awarded in respect of “an employment practice that is unlawful because of its disparate impact.” The district court denied certification, and the plaintiffs asked this court for leave to appeal the denial. A motions panel granted leave, but the defendant argues that the panel erred — that the appeal is untimely. We begin with that question. Rule 23(

f) of the civil rules permits appeals from orders granting or denying class certification despite the general policy (though one with many exceptions) against allowing interlocutory appeals in the federal court system. A denial of class certification often dooms the suit — the class members’ claims may be too slight to justify the expense of individual suits. Conversely, because of the astronomical damages potential of many class action suits, a grant of certification may place enormous pressure on the defendant to settle even if the suit has little merit. See, e.g., CE Design Ltd. v.

King Architectural Metals, Inc., 637 F.3d 721, 723 (7th Cir. 2011). And because class actions are cumbersome and protracted, an early appellate decision on whether a suit can be maintained as a class action can speed the way to termination of the litigation by abandonment,

summary judgment, or settlement. E.g., Blair v. Equifax Check Services, Inc., 181 F.3d 832, 834-35 (7th Cir.1999); Newton v. Merrill Lynch, 259 F.3d 154, 162-65 (3d Cir.2001). But Rule 23(

f) requires that leave to appeal be sought from the court of appeals within 14 days of the entry of the order granting or denying certification. The district court denied the plaintiffs’ initial motion for class certification in August 2010. In July 2011 the plaintiffs filed an amended motion for class certification, which the district judge denied in September, and within 14 days of that denial the plaintiffs sought our leave to appeal. The defendant asks us to treat the request for leave to appeal as an untimely request to appeal the August 2010 denial of certification.

That would amount to treating the plaintiffs’ second motion for certification as an untimely motion to reconsider the denial of the first motion. The question of timeliness may seem to be about jurisdiction, since most deadlines for appeals from a district court have been held to be jurisdictional. But as we noted recently in In re IFC Credit Corp., 663 F.3d 315, 319-20 (7th Cir.2011), the Supreme Court has been moving toward a definition of the subject-matter jurisdiction of the federal courts that includes all cases that these courts are “competent,” in the sense of legally empowered, to decide.

This implies that deadlines for appealing are not jurisdictional, since they regulate the movement upward through the judicial hierarchy of litigation that by definition is within federal jurisdiction. Yet appeal deadlines either found in statutes or adopted by courts by direction of a statute continue to be treated as jurisdictional— though not all of them; the Supreme Court recently rejected such a “bright line” rule in favor of requiring a “clear indication” that the deadline was intended by Congress to be jurisdictional. Henderson v.

Shinseki, — U.S.-, 131 S.Ct. 1197, 1203, 179 L.Ed.2d 159 (2011). (The power of Congress to impose such limits on the jurisdiction of the federal courts is not questioned.) But because no “clear indication” is to be found in the pertinent statutory texts, see, e.g., 28 U.S.C. §§ 2101(c), 2107(a), (c), the Court has found itself saying such things as that Congress is not required to “use magic words in order to speak clearly on this point” and that “context, including [the Supreme Court’s]

interpretation of similar provisions in many years past, is relevant.” Henderson v. Shinseki, supra, 131 S.Ct. at 1203, quoting Reed Elsevier, Inc. v. Muchnick, — U.S.-, 130 S.Ct. 1237, 176 L.Ed.2d 18 (2010). What we take away from this formula is that if the Court has traditionally treated a particular statutory deadline as jurisdictional it will go on doing so, id. at 1203-06; John R. Sand & Gravel Co. v. United States, 552 U.S. 130, 134, 128 S.Ct. 750, 169 L.Ed.2d 591 (2008); Bowles v.

Russell, 551 U.S. 205, 209-10 and n. 2, 127 S.Ct. 2360, 168 L.Ed.2d 96 (2007); In re Caterbone, 640 F.3d 108, 111-13 (3d Cir.2011), even though doing so doesn’t comport with the new “competence” standard. Deadlines for appealing are just a type of statute of limitations, as acknowledged in John R. Sand & Gravel v. United States, supra, 552 U.S. at 133, 128 S.Ct. 750, and statutes of limitations ordinarily are affirmative defenses rather than jurisdictional bars. A deadline for bringing or appealing a federal case presupposes that the case is within the competence of federal courts to decide.

We declined in Asher v. Baxter Int’l Inc., 505 F.3d 736, 741 (7th Cir.2007), to rule on whether the deadline in Rule 23(f), though it is promulgated by the Supreme Court under the authority of the Rules Enabling Act, 28 U.S.C. § 2072, rather than found in or directed to be adopted by a statute, is jurisdictional. But by now it is clear that it is not jurisdictional — that the exception to the “competence” standard is limited to statutory deadlines, United States v.

Neff, 598 F.3d 320, 322-23 (7th Cir.2010), for how can a court contract or expand its jurisdiction except by force of a constitutional or statutory provision? If the deadline was made by Congress, then whether it is jurisdictional depends on congressional intent, and, the Supreme Court appears to be saying, in the absence of any clues to that intent on whether the courts traditionally have treated the deadline as jurisdictional.

The time limit in Rule 23(f), having been created by the Court rather than by Congress (no time limits are specified in the Rules Enabling Act — the Act is an enabler, not a specifier), is governed by the “competence” standard and therefore is not jurisdictional, for obviously the suit from which the appeal is sought to be taken is within the jurisdiction of the federal courts. But suppose our understanding of the evolving Supreme Court doctrine is wrong, and the deadline in Rule 23(

f) is jurisdictional. The only difference between a deadline that is jurisdictional and one that is not is that a litigant cannot lose the benefit of the former type (until judgment becomes final after exhaustion of appellate remedies) by failing to assert it, or because the other party’s failure to comply would in nonjurisdictional settings be excused by such doctrines as equitable estoppel or equitable tolling. The defendant has from the outset vigorously contested the timeliness of the appeal, and the plaintiffs are not arguing that they should be excused for having missed the deadline.

Even if not jurisdictional, a deadline is mandatory in the sense that if invoked by a party in timely fashion the court is bound by it. Eberhart v. United States, 546 U.S. 12, 19, 126 S.Ct. 403, 163 L.Ed.2d 14 (2005) (per curiam); Asher v. Baxter Int’l Inc., supra, 505 F.3d at 741; Maxwell v. Dodd, 662 F.3d 418, 421 (6th Cir.2011); Wilburn v. Robinson, 480 F.3d 1140, 1146-47 (D.C.Cir.2007). Rather, the plaintiffs’ argument is that their 14 days to seek leave to appeal ran anew from the denial of their amended motion for class certification.

The defendant points out that a deadline for appealing cannot be extended by a motion for reconsideration of a previous appealable order, Asher v. Baxter Int’l Inc., supra, 505 F.3d at 739-40; Gary v. Sheahan, 188 F.3d 891 (7th Cir.1999); Jenkins v. Bell-South Corp., 491 F.3d 1288, 1290-92 (11th Cir.2007); McNamara v. Felderhof, 410 F.3d 277, 280-81 and n. 8 (5th Cir.2005), unless the motion is made within the time allowed for taking the appeal, Blair v. Equifax Check Services, Inc., supra, 181 F.3d at 837, and this rule applies to appeals under Rule 23(f)- Id.

Otherwise the deadline for taking the appeal would be eviscerated. And this is so even if the motion for reconsideration doesn’t just say “and for the reasons stated in our original motion we ask the court to reverse its ruling” but adds “and by the way we have thought of some clever new arguments for why our motion should have been granted.” Carpenter v. Boeing Co., 456 F.3d 1183, 1190-91 (10th Cir.2006). For it is easy to think up new arguments.

But it doesn’t follow that the failure to take a timely appeal from one interlocutory order operates as a forfeiture, jurisdictional or otherwise, of the right to appeal a subsequent order. For the later motion may not be, either in form or, more important, in substance, a motion to reconsider the previous denial. A rule limiting parties to one interlocutory appeal from a grant or denial of class certification would disserve Rule 23(f). It is important that the question whether the case is to proceed as a class action be resolved sooner rather than later.

So if it becomes clear in the course of the lawsuit, as a result of new law or newly learned facts, that the denial of certification was erroneous, and if years of litigation lie ahead before a final judgment can be expected, and if therefore an appeal from the denial of certification may either end the litigation or at least place it on a path to swift resolution, the court of appeals should have discretion to allow the appeal. The fact that the appellate court has a discretionary jurisdiction over Rule 23(

f) appeals is important. Appellate jurisdiction in the federal system ordinarily is mandatory. With few exceptions, we have to decide all appeals that we have jurisdiction to hear; we do not have a discretionary appellate jurisdiction like the Supreme Court. But because our jurisdiction to hear interlocutory appeals under Rule 23(

f) is discretionary, there is little danger that the filing in the district court of a second motion for certification based on altered circumstances, followed if it is denied by a motion in this court for leave to appeal, will either delay the district court proceedings (Rule 23(

f) provides that “an appeal does not stay proceedings in the district court unless the district judge or the court of appeals so orders”) or burden us or the opposing party. If the movant is playing a delay game, prompt denial by the motions panel of leave to appeal probably will end it; if he persists he will be courting sanctions in both the district court and this court for filing frivolous pleadings. And by the way, we do not permit a party to circumvent the 14-day deadline in Rule 23(

f) by appealing a denial of class certification under 28 U.S.C. § 1292(b) (authorizing interlocutory appeals that present a controlling issue of law on which there is substantial room for disagreement and prompt resolution would expedite the litigation), Richardson Electronics, Ltd. v. Panache Broadcasting, 202 F.3d 957, 959 (7th Cir.2000), which has no deadline. Dicta in the Tenth Circuit’s opinion in Carpenter v.

Boeing Co., supra, go beyond the unexceptionable proposition that merely presenting “new arguments” does not change a motion for reconsideration of a grant or denial of class certification into a motion that if denied is appealable under Rule 23(f). The opinion states (456 F.3d at 1191) that given the multifactor analysis that courts must apply in deciding the propriety of class certification, [even appellate review limited to whatever changed circumstances had given rise to the fresh motion for certification] would often require contorted thinking that exceeds the capacities of even appellate courts.

How can an appellate court say that one particular new factor would require a different result regardless of how the district court weighed the factors presented originally? In stating that the new factor required a different result, the appellate court must engage in weighing the factors weighed by the district court in its original ruling but cannot know precisely how much weight the district court granted to each. In particular, what if the district court clearly erred in giving dispositive weight to one factor?

How is the appellate court to ignore such error (in keeping with the presumption that the original decision was correct) even when it addresses a motion for reconsideration that raises only a rather inconsequential new factor? ... We are not inclined to adopt a construction of Rule 23(

f) that would regularly require mental gymnastics just for the purpose of giving litigants a second bite at the interlocutory-appellate-review apple. We note that the very absence of a prompt appeal by the party aggrieved by the decision on certification suggests that the concerns justifying Rule 23(

f) are, at the least, less significant in the particular case. If the decision whether or not to certify the class was truly outcome determinative, one would not expect the losing party to continue the litigation for months before launching a new challenge to the ruling. Any value in permitting a belated interlocutory appeal is overridden by the desirability of the district court’s proceeding expeditiously. A court of appeals is never obliged to engage in “contorted thinking” about a Rule 23(

f) appeal, for it can always deny leave to appeal, and should do so if it would have to do mental contortions in order to make up its collective mind whether appeal should be allowed.

And if the new motion for certification “raises only a rather inconsequential new factor,” then the failure of the plaintiffs to have sought interlocutory review of the denial of the original motion for certification becomes a reason to deny leave to appeal out of hand, without any “mental gymnastics.” As for “not expecting] the losing party to continue the litigation for months before launching a new challenge to the ruling,” the new challenge is timely if filed as soon as the development warranting a new motion for certification occurs, but untimely if the plaintiff dawdles.

And if the appeal is not “belated,” but based on developments that may warrant certification, allowing the appeal may very well speed up rather than slow down the litigation. In effect the court held in Carpenter that the new motion for certification was in substance an untimely motion for reconsideration. The holding is unexceptionable, but the dicta are not persuasive. The basis of the plaintiffs’ renewed motion for class certification in the present case was the Supreme Court’s decision in Wal-Mart Stores, Inc. v. Dukes, —U.S.-, 131 S.Ct. 2541, 180 L.Ed.2d 374 (2011), handed down a month earlier.

That was an important development in the law governing class certification in employment discrimination cases — possibly a milestone. It may seem a perverse basis for a renewed motion for class certification, since the Supreme Court reversed a grant of certification in what the defendant in our case insists is a case just like this one. But the district judge, though he again denied certification, didn’t think the plaintiffs were perverse in basing their new motion on Wal-Mart. On the contrary, he said that “Wal-Mart does add a lot to the landscape under Rule 23----I think this really cries out for a 23(

f) appeal, and I would support it. And I’m going to put that in my order [denying the renewed motion for certification — and he did]____ [T]his is one [case] that really cries out for [a Rule 23(

f) appeal] with the change in the landscape by the Wal-Mart opinion. Of course, you know, most defendants think that the change is in their interest, not in the plaintiffs’. But you’ve [the judge was addressing the plaintiffs’ lawyer] made a good argument, and I think it deserves to be put to rest one way or the other.” The judge was right.

Wal-Mart holds that if employment discrimination is practiced by the employing company’s local managers, exercising discretion granted them by top management (granted them as a matter of necessity, in Wal-Mart’s case, because the company has 1.4 million U.S. employees), rather than implementing a uniform policy established by top management to govern the local managers, a class action by more than a million current and former employees is unmanageable; the incidents of discrimination complained of do not present a common issue that could be resolved efficiently in a single proceeding. Fed.R.Civ.P. 23(a)(2).

Not that the employer would be immune from liability even in such a case; if the local managers are acting within the scope of their employment in discriminating against their underlings on a forbidden ground (sex, alleged in Wal-Mart, race in our case), the employer is liable for their unlawful conduct under the doctrine of respondeat superior.

But because there was no company-wide policy to challenge in Wal-Mart — the only relevant corporate policies were a policy forbidding sex discrimination and a policy of delegating employment decisions to local managers— there was no common issue to justify class treatment.

The district judge thought this case like Wal-Mart because Merrill Lynch, accused of discriminating against 700 black brokers currently or formerly employed by it, delegates discretion over decisions that influence the compensation of all the company’s 15,000 brokers (“Financial Advisors” is their official title) to 135 “Complex Directors.” Each of the Complex Directors supervises several of the company’s 600 branch offices, and within each branch office the brokers exercise a good deal of autonomy, though only within a framework established by the company.

Two elements of that framework are challenged: the company’s “teaming” policy and its “account distribution” policy. The teaming policy permits brokers in the same office to form teams. They are not required to form or join teams, and many prefer to work by themselves. But many others prefer to work as part of a team.

Team members share clients, and the aim in forming or joining a team is to gain access to additional clients, or if one is already rich in clients to share some of them with brokers who have complementary skills that will secure the clients’ loyalty and maybe persuade them to invest more with Merrill Lynch. As we said, there are lone wolves, but there is no doubt that for many brokers team membership is a plus; certainly the plaintiffs think so. The teams are formed by brokers, and once formed a team decides whom to admit as a new member.

Complex Directors and branch-office managers do not select the team’s members. Account distributions are transfers of customers’ accounts when a broker leaves Merrill Lynch and his clients’ accounts must therefore be transferred to other brokers. Accounts are transferred within a branch office, and the brokers in that office compete for the accounts. The company establishes criteria for deciding who will win the competition. The criteria include the competing brokers’ records of revenue generated for the company and of the number and investments of clients retained.

The Complex Directors, as well as the branch-office managers, have a measure of discretion with regard to teaming and account distribution; they can veto teams and can supplement the company criteria for distributions. And to the extent that these regional and local managers exercise discretion regarding the compensation of the brokers whom they supervise, the case is indeed like Wal-Mart.

But the exercise of that discretion is influenced by the two company-wide policies at issue: authorization to brokers, rather than managers, to form and staff teams; and basing account distributions on the past success of the brokers who are competing for the transfers. Furthermore, team participation and account distribution can affect a broker’s performance evaluation, which under company policy influences the broker’s pay and promotion. The plaintiffs argue that these company-wide policies exacerbate racial discrimination by brokers.

The teams, they say, are little fraternities (our term but their meaning), and as in fraternities the brokers choose as team members people who are like themselves. If they are white, they, or some of them anyway, are more comfortable teaming with other white brokers. Obviously they have their eyes on the bottom line; they will join a team only if they think it will result in their getting paid more, and they would doubtless ask a superstar broker to join their team regardless of his or her race.

But there is bound to be uncertainty about who will be effective in bringing and keeping shared clients; and when there is uncertainty people tend to base decisions on emotions and preconceptions, for want of objective criteria. Suppose a police department authorizes each police officer to select an officer junior to him to be his partner. And suppose it turns out that male police officers never select female officers as their partners and white officers never select black officers as their partners.

There would be no intentional discrimination at the departmental level, but the practice of allowing police officers to choose their partners could be challenged as enabling sexual and racial discrimination — as having in the jargon of discrimination law a “disparate impact” on a protected group — and if a discriminatory effect was proved, then to avoid an adverse judgment the department would have to prove that the policy was essential to the department’s mission. 42 U.S.C. § 2000e-2(k)(1)(A)(i); Ricci v. DeStefano, 557 U.S. 557, 129 S.Ct. 2658, 2672-73, 174 L.Ed.2d 490 (2009); Bryant v.

City of Chicago, 200 F.3d 1092, 1098-99 (7th Cir.2000). That case would not be controlled by Wal-Mart (although there is an undoubted resemblance), in which employment decisions were delegated to local managers; it would be an employment decision by top management. Merrill Lynch’s broker teams are formed by brokers, not managers, just as in our hypothetical example police officers’ partners are chosen by police officers, not supervisors.

If the teaming policy causes racial discrimination and is not justified by business necessity, then it violates Title VII as “disparate impact” employment discrimination — and whether it causes racial discrimination and whether it nonetheless is justified by business necessity are issues common to the entire class and therefore appropriate for class-wide determination.

And likewise with regard to account distributions: if as a result of racial preference at the team level black brokers employed by Merrill Lynch find it hard to join teams, or at least good teams, and as a result don’t generate as much revenue or attract and retain as many clients as white brokers do, then they will not do well in the competition for account distributions either; and a kind of vicious cycle will set in.

A portion of a team’s pre-existing revenues are transferred within a team to a new recruit, who thus starts out with that much “new” revenue credited to him or her — an advantage, over anyone who is not on a team and thus must generate all of his own “new” revenue, that translates into a larger share of account distributions, which in turn helps the broker do well in the next round of such distributions.

This spiral effect attributable to company-wide policy and arguably disadvantageous to black brokers presents another question common to the class, along with the question whether, if the team-inflected account distribution system does have this disparate impact, it nevertheless is justified by business necessity. There is no indication that the corporate level of Merrill Lynch (or its parent, Bank of America) wants to discriminate against black brokers. Probably it just wants to maximize profits.

But in a disparate impact case the presence or absence of discriminatory intent is irrelevant; and permitting brokers to form their own teams and prescribing criteria for account distributions that favor the already successful— those who may owe their success to having been invited to join a successful or promising team — are practices of Merrill Lynch, rather than practices that local managers can choose or not at their whim.

Therefore challenging those policies in a class action is not forbidden by the Wal-Mart decision; rather that decision helps (as the district judge sensed) to show on which side of the line that separates a company-wide practice from an exercise of discretion by local managers this case falls. Echoing the district judge, the defendant’s brief states that “any discrimination here would result from local, highly-individualized implementation of policies rather than the policies themselves.” That is too stark a dichotomy.

Assume that with no company-wide policy on teaming or account distribution, but instead delegation to local management of the decision whether to allow teaming and the criteria for account distribution, there would be racial discrimination by brokers or local managers, like the discrimination alleged in WalMart. But assume further that company-wide policies authorizing broker-initiated teaming, and basing account distributions on past success, increase the amount of discrimination.

The incremental causal effect (overlooked by the district judge) of those company-wide policies — which is the alleged disparate impact — could be most efficiently determined on a class-wide basis. We are not suggesting that there is in fact racial discrimination at any level within Merrill Lynch, or that management’s teaming and account distribution policies have a racial effect. The fact that black brokers have on average lower earnings than white brokers may have different causes altogether.

The only issue at this stage is whether the plaintiffs’ claim of disparate impact is most efficiently determined on a class-wide basis rather than in 700 individual lawsuits. The district judge exaggerated the impact on the feasibility and desirability of class action treatment of the fact that the exercise of discretion at the local level is undoubtedly a factor in the differential success of brokers, even if not a factor that overwhelms the effect of the corporate policies on teaming and on account distributions.

Obviously a single proceeding, while it might result in an injunction, could not resolve class members’ claims. Each class member would have to prove that his compensation had been adversely affected by the corporate policies, and by how much. So should the claim of disparate impact prevail in the class-wide proceeding, hundreds of separate trials may be necessary to determine which class members were actually adversely affected by one or both of the practices and if so what loss each class member sustained — and remember that the class has 700 members.

But at least it wouldn’t be necessary in each of those trials to determine whether the challenged practices were unlawful. Rule 23(c)(4) provides that “when appropriate, an action may be brought or maintained as a class action with respect to particular issues.” The practices challenged in this case present a pair of issues that can most efficiently be determined on a class-wide basis, consistent with the rule just quoted. As said in Mejdrech v.

Met-Coil Systems Corp., 319 F.3d 910, 911 (7th Cir. 2003), class action treatment is appropriate and is permitted by Rule 23 when the judicial economy from consolidation of separate claims outweighs any concern with possible inaccuracies from their being lumped together in a single proceeding for decision by a single judge or jury. Often, and as it seems to us here, these competing considerations can be reconciled in a “mass tort” case by carving at the joints of the parties’ dispute.

If there are genuinely common issues, issues identical across all the claimants, issues moreover the accuracy of the resolution of which is unlikely to be enhanced by repeated proceedings, then it makes good sense, especially when the class is large, to resolve those issues in one fell swoop while leaving the remaining, claimant-specific issues to individual follow-on proceedings.

The kicker is whether “the accuracy of the resolution” would be “unlikely to be enhanced by repeated proceedings.” If resisting a class action requires betting one’s company on a single jury verdict, a defendant may be forced to settle; and this is an argument against definitively resolving an issue in a single case if enormous consequences ride on that resolution. In re Bridgestone/Firestone, Inc., 288 F.3d 1012, 1020 (7th Cir.2002); In re Rhone-Poulenc Rorer, Inc., 51 F.3d 1293, 1299-1300 (7th Cir.1995); contra, Klay v. Humana, Inc., 382 F.3d 1241, 1274 (11th Cir.2004).

But Merrill Lynch is in no danger of being destroyed by a binding class-wide determination that it has committed disparate impact discrimination against 700 brokers, although an erroneous injunction against its teaming and account distribution policies could disadvantage it in competition with brokerage firms that employ similar policies — though we have no information on whether others do.

The Mejdrech decision, and Bridge-stone/Firestone and Rhone-Poulenc more fully, discuss the danger that resolving an issue common to hundreds of different claimants in a single proceeding may make too much turn on the decision of a single, fallible judge or jury. The alternative is multiple proceedings before different triers of fact, from which a consensus might emerge; a larger sample provides a more robust basis for an inference. But that is an argument for separate trials on pecuniary relief, and the only issue of relief at present is whether to allow the plaintiffs to seek class-wide injunctive relief.

There isn’t any feasible method — certainly none has been proposed in this case — for withholding injunctive relief until a series of separate injunctive actions has yielded a consensus for or against the plaintiffs. As far as pecuniary relief is concerned, there may be no common issues (though then again there may be, see Allen v.

International Truck & Engine Corp., supra, 358 F.3d at 472), and in that event the next stage of the litigation, should the class-wide issue be resolved in favor of the plaintiffs, will be hundreds of separate suits for backpay (or conceivably for compensatory damages and even punitive damages as well, if the plaintiffs augment their disparate-impact claim with proof of intentional discrimination). The stakes in each of the plaintiffs’ claims are great enough to make individual suits feasible.

Most of Merrill Lynch’s brokers earn at least $100,000 a year, and many earn much more, and the individual claims involve multiple years. But the lawsuits will be more complex if, until issue or claim preclusion sets in, the question whether Merrill Lynch has violated the antidiscrimination statutes must be determined anew in each case. We have trouble seeing the downside of the limited class action treatment that we think would be appropriate in this case, and we conclude that the district judge erred in deciding to the contrary (with evident misgivings, however).

The denial of class certification under Rules 23(b)(2) and (c)(4) is therefore Reversed.