United States District Court, N.D. Illinois, Eastern Division
September 12, 2014
MICHAEL J. CHRISTIAN, BETY FLEMMING, EARTHA C. JOHNSON, as Attorney-in-Fact for MINNIE ALICE LARK, and CLAUDIA DORTCH, on behalf of themselves and all others similarly situated, Plaintiffs,
GENERATION MORTGAGE COMPANY, Defendant.
JOHN J. THARP, Jr., District Judge.
Plaintiffs have sought reconsideration of the Court's order granting Generation Mortgage Company's motion to bifurcate class and merits discovery in this matter. Dkt. 83 ("Order"). That motion was predicated on the plaintiffs' efforts to obtain data from tens of thousands of individual loan files for use in a statistical analysis comparing the loan terms offered to members of the putative classes (African Americans and single females) to those outside those classes. Generation argued that it should not be required to undergo such hugely burdensome and expensive class-wide discovery unless and until the plaintiffs have obtained certification of the putative classes. The plaintiffs argue that they need class-wide discovery in order to make the showing of commonality necessary to obtain class certification.
Although framed as a discovery dispute, at its core this dispute involves the relevance of the statistical analysis the plaintiffs wish to conduct with Generation's loan data.Relying on Wal-Mart Stores, Inc. v. Dukes, 131 S.Ct. 2541 (2011), the Court concluded that the proposed regression analysis would not suffice to establish that Generation's mortgage brokers exercised their pricing discretion in a discriminatory manner because "a statistical disparity in the loan terms received by African-Americans and/or single women says nothing about the cause of the disparity." Order at 4. The Court therefore granted Generation's bifurcation motion, limited initial discovery in the case to evidence pertinent to the question of whether Generation's brokers exercised their pricing discretion in a common manner, and held that the statistical evidence needed for the plaintiffs' proposed regression analysis is not relevant to that question.
Plaintiffs' primary argument for reconsideration asserts that the Court erroneously assumed that "plaintiffs' statistical analysis of loan data will not control for all the factors that may account for any observed disparities in interest rates and origination fees." If by "may account for" the plaintiffs mean "cause, " it is true enough that the Court drew that conclusion- for two substantial reasons. First, the plaintiffs conceded the point in their response. And second, the Supreme Court's opinion in Wal-Mart compels the conclusion.
In their response to Generation's motion to bifurcate, the plaintiffs acknowledged that their complaint " does not assert that regression analysis alone will resolve the class or commonality questions "; rather, "statistical analysis of loans will uncover whether this policy had a common effect. " Resp., Dkt. 73, at 14 (emphasis added). See also id. at 1 and 12 (statistical analysis necessary but "not sufficient" to show commonality). That is because the data the plaintiffs seek will, at best, establish "the common effect of Generation's nationwide policy." Surreply, Dkt. 80, at 2 (emphasis added).
A common effect does not establish a common cause for that effect, even in a disparate impact case. That is the Supreme Court's teaching in Wal-Mart; which "reiterates that statistical correlation, no matter how robust, cannot substitute for a specific finding of class-action commonality." In re Countrywide Financial Corp. Mortgage Lending Practices Litig., 708 F.3d 704, 709 (6th Cir. 2013) (citing Wal-Mart, 131 S.Ct. at 2555, and affirming the denial of certification of class of borrowers because plaintiffs statistical evidence of disparity was insufficient to establish that the disparity was the product of a single cause). As the plaintiffs have acknowledged, disparate impact is a necessary but not sufficient condition to establish their claim.
In the context of a class disparate impact claim, the plaintiffs must identify a "specific employment practice" that is the common cause of the disparate impact. Wal-Mart, 131 S.Ct. at 2555. As noted in this Court's original opinion, the plaintiffs attempted to do this by attributing the alleged disparate impact to Generation's policy of affording pricing discretion to its brokers, but Wal-Mart squarely rejected the proposition that a company's policy of permitting discretion to managers can serve as the common link between individual claims in a disparate impact case. Id. at 2554-55. The "bare existence of delegated discretion, " id. at 2555, is no policy at all-or, viewed another way, it is a policy against having a policy. Id. at 2554. Wal-Mart holds that showing that such a non-policy produces disparate impact "does not suffice" to satisfy Rule 23's commonality requirement. Id. at 2556.
Notwithstanding this unequivocal holding, the plaintiffs continue to argue that Generation's policy of pricing discretion does suffice as the common cause of the alleged disparate impact because that policy cabins the discretion afforded to brokers by setting a minimum rate. Generation's brokers, in other words, only have discretion to increase the rate, not to lower it, and because they receive more compensation for higher rates, the plaintiffs argue that the policy incentivizes brokers to increase rates above the minimum approved rate. The Court rejected this purported distinction as illusory;the rationale for any company policy-including the policy of giving hiring and promotion discretion to local managers that was at issue in Wal-Mart -presumably can be linked to the company's desire to increase profits. That does not change the fact that in affording discretion to managers, or here, brokers, the company replaces centralized decisionmaking with localized decisionmaking. And in that context, as Wal-Mart holds, there is no "glue" to tie those individualized decisions together in a class action.
Plaintiffs argue that they "are not required to pinpoint exactly which business practice" caused a disparate impact, Supp. Br., Dkt. 106 at 4-5; they cite 42 U.S.C § 2000e-2(k)(1)(B)(i) for the proposition that "the elements of a decision-making process may not be capable of separation for analysis, in which case it may be appropriate to challenge the decision-making process as a whole." Id. But that argument merely rephrases the proposition that Wal-Mart rejected, namely that it suffices to identify a general policy of managerial discretion as the common cause of a disparate impact on a protected group. 131 S.Ct. at 2555-56 (rejecting the identification of a policy of delegated discretion as sufficient identification of a specific policy on which to base a disparate impact claim). As the Supreme Court observed in Watson v. Fort Worth Bank & Trust, 487 U.S. 977, 990 (1988), and reiterated in Wal-Mart, 131 S.Ct. at 2554, a policy of permitting managerial discretion "should itself raise no inference of discriminatory conduct." Accordingly, merely lumping together an amalgamation of subjective individual decisions does not satisfy this standard. See, e.g., Davis v. Cintas Corp., 717 F.3d 476, 497 (6th Cir. 2013) (affirming summary judgment where plaintiff failed to identify the challenged employment practice with more particularity than aggregating all of the subjective elements of a hiring system). The Seventh Circuit has similarly left no room for doubt on this score:
For disparate impact claims, a plaintiff must establish that a particular employment practice causes a disparate impact on a member of a protected class. To satisfy this burden, the plaintiff is responsible for isolating and identifying the specific employments [sic] practices that are allegedly responsible for any observed statistical disparities. Notably, it is not enough to simply allege that there is a disparate impact on workers, or point to a generalized policy that leads to such an impact. Failure to identify the specific practice could lead to employers being held liable for the myriad of innocent causes that may lead to statistical imbalances.
Pell v. Allstate Ins. Co., 675 F.3d 709, 717 (7th Cir. 2012) (emphasis added).
In advancing this argument, moreover, the plaintiffs fail to distinguish between the requirements for an individual disparate impact claim and the additional requirements for certification of a class. As the Court explained in Wal-Mart, that an individual may advance a disparate impact claim based on the existence of a policy of discretion "does not lead to the conclusion that every employee in a company using a system of discretion has such a claim in common." 131 S.Ct. at 2554. What is required, rather, is the identification of "a common mode of exercising discretion that pervades the entire company." Id. at 2554-55. Given the pricing discretion afforded to Generation's brokers, the putative class members must focus on the implementation of that discretion by the brokers; to tie their claims together in a class, they are required to show (among other things) that the brokers exercised their discretion in a common, discriminatory, manner.
Which brings us back to the question of the utility of the plaintiffs' proposed regression analysis. As noted, the plaintiffs conceded in their response and surreply to the defendant's bifurcation motion that their proffered statistical analysis does not suffice to establish commonality: plaintiffs do " not assert that regression analysis alone will resolve the class or commonality questions." Plt's Amended Response (Dkt. 73) at 14 (emphasis in original); see also id. at 1 and 12 (statistical analysis "not sufficient" to show commonality). That concession should end the debate, because the plaintiffs also concede that they have no other means to establish commonality:
The only possible way to know whether Generation's brokers exercised their pricing discretion in a common direction is for Generation to produce the evidence of each pricing decision of each broker for each borrower in the class period. There simply is no "other means" in existence for discovering and showing "whether Generation's brokers exercised their pricing discretion in a common manner." By making the well-intentioned but uninformed suggestion that plaintiffs use "other means" to discover and establish the point, the Court is, in effect challenging plaintiffs to drive a car from which it has removed the wheels.
Plt's Mem. in Support of Motion for Reconsideration, Dkt. 89 at 6-7 (quoting Order).
Despite their initial and repeated concessions that statistical evidence does not suffice to prove causation, the plaintiffs argue inconsistently in their motion for reconsideration that the Court "improperly assumes... that plaintiffs' statistical analysis will not effectively isolate and demonstrate that the cause of any disparities in interest rates and origination fees is racial or gender discrimination." Id. at 2. Now, the plaintiffs maintain, statistical analysis does suffice to prove causation. Confronting the implications of their concession, the plaintiffs have simply changed their tune.
Even if reversing course were a permissible basis on which to seek reconsideration (and it is not), the argument flies in the face of the Court's opinion in Wal-Mart. Simply put, Wal-Mart held that the identification of "potential" discrimination through statistical analysis is insufficient to satisfy Rule 23(a)'s commonality requirement where the decision makers are vested with independent discretion. The Court expressly rejected as "insufficient to establish that respondents' theory can be proved on a classwide basis, " 131 S.Ct. at 2555, not one, but two, regression analyses that the plaintiffs had commissioned. The Court explained that even if the analyses established commonalities at the national or regional level (as they purported to do), they would still fail to "establish the existence of disparities at individual stores, let alone raise the inference that a company-wide policy of discrimination is implemented by discretionary decisions at the store and district level." Id.
The plaintiffs argue that Wal-Mart did not categorically reject the possibility that a sufficiently thorough regression analysis could suffice to establish commonality and maintain that the regression analysis that their expert will conduct will be better than those the Court deemed insufficient in Wal-Mart. But they are wrong as to the first predicate, making the second irrelevant. In rejecting the plaintiffs' proffered statistical evidence, the Supreme Court explained:
There is another, more fundamental, respect in which respondents' statistical proof fails. Even if it established (as it does not) a pay or promotion pattern that differs from the nationwide figures or the regional figures in all of Wal-Mart's 3, 400 stores, that would still not demonstrate that commonality of issue exists.... [M]erely proving that the discretionary system has produced a racial or sexual disparity is not enough. ... Other than the bare existence of delegated discretion, respondents have identified no specific employment practice'-much less one that ties all their 1.5 million claims together. Merely showing that Wal-Mart's policy of discretion has produced an overall sex-based disparity does not suffice.
131 S.Ct. at 2555-56 (emphasis in original).
This Court directed the parties to present supplemental briefs on "the meaning and relevance of the Supreme Court's discussion in Wal-Mart of the more fundamental respect in which respondents' statistical proof fails, set forth at 131 S.Ct. 2555-56." Dkt. 95. Remarkably, the plaintiffs contend that the import of this passage is that the statistical analysis offered in Wal-Mart "merely identified aggregate national and regional disparities." Plt's Supplemental Brief, Dkt. 106, at 2. According to the plaintiffs, it was critical to the Court's analysis in Wal-Mart that "the plaintiffs did not even attempt to establish disparities at Walmart's individual stores or to control for the possibility that regional disparities could be driven by just a few stores." Id.
The plaintiffs' reading cannot be squared with the Supreme Court's language or analysis. While the Court noted the failure of the plaintiffs to account for potential regional and store-by-store disparities, that was not the "fundamental" problem with the regression analyses they offered. "More fundamental, " the Court said, was the inability to show by statistical evidence that all of the discretionary decisions at issue had been the result of a common mode of exercising discretion by all of those in whom that discretion was vested. 131 S.Ct. at 2555. The Supreme Court expressly rejected the argument, urged by the plaintiffs here, that had the regression analysis been sufficiently comprehensive to account for disparities on even a store-by-store basis, it would have passed muster: "Even if it established... a pay or promotion pattern that differs from the nationwide figures or the regional figures in all of Wal-Mart's 3, 400 stores, that would still not demonstrate that commonality of issue exists. " Id. (emphasis added). Even then, the Court concluded, the statistical evidence would show only that the policy of discretion had produced a sex-based disparity, not that every observed disparity (whether one national disparity or 3400 store-by-store disparities) was the product of a common mode of exercising discretion. In other words, the Supreme Court concluded-as did this Court in its original opinion-that a statistical disparity showing a disparate impact (in Wal-Mart, fewer promotions and lower pay; here, less favorable loan terms) is simply inadequate to establish commonality when the "policy" under review is a general policy that affords widespread discretion over the decisions at issue-that is, "a policy against having uniform practices." 131 S.Ct. at 2554.
In short, the "fundamental" problem with the arguments advanced by the Wal-Mart plaintiffs and reasserted by the plaintiffs here, is that there is no common practice to analyze. Statistical analysis-no matter how rigorous-cannot solve the commonality problem when every decision under review is discretionary. This is not a question of the statistician's skill but of the supervisor's subjectivity. A policy of discretion allows each supervisor, or here, broker, to base pricing decisions on a unique-not common-combination of subjective and objective factors. By definition, in this context, there can be no "common" practice.
It is perhaps unsurprising, then, that the plaintiffs do not identify a single post- Wal-Mart case in which a court has determined that any statistical analysis has sufficed to establish common causation where a company vests managers (or here, brokers) with independent discretion to make the allegedly discriminatory decisions.To the contrary, it appears (as defendant contends) that courts considering such arguments have uniformly rejected them. The Seventh Circuit has plainly held that "when multiple managers exercise independent discretion, conditions at different stores (or sites) do not present a common question." Bolden v. Walsh Constr. Co., 688 F.3d 893, 896 (7th Cir. 2012). See also Vang v. Kohler Co., 488 F.Appx. 146, 147 (7th Cir. 2012) (remanding class certification decision for reconsideration in light of Wal-Mart and Bolden; "Under Wal-Mart and Bolden such variable circumstances do not present a common question.").
More recently, in cases substantially similar to this one, the Sixth and Third Circuits affirmed district court decisions premised on Wal-Mart' s holding that statistical correlation cannot establish commonality of cause for a disparate impact when loan officers operate under a policy of pricing discretion. In Countrywide Financial Corp. Mortgage Lending Practices Litig., the Sixth Circuit affirmed the district court's denial of class certification in a case in which the putative class claimed that Countrywide's loan pricing policy, which permitted its local agents broad discretion to depart from par rates that had been set solely by reference to objective factors. 708 F.3d at 706. The plaintiffs presented statistical evidence to support their claim "that vesting local agents with discretion to deviate from objective par rates led Countrywide" to charge minority borrowers rates that were 11-12 basis points higher than the rates paid by whites. Noting the similarity between the claims of the plaintiffs in the two cases, the Sixth Circuit concluded that although statistical analysis remains relevant to proving disparate impact, it has no role to play in establishing commonality where the policy at issue is a policy that affords substantial discretion: " Wal-Mart reiterates that statistical correlation, no matter how robust, cannot substitute for a specific finding of class-action commonality." Id. at 709. See also Davis, 717 F.3d at 488-89 (affirming summary judgment denying disparate impact claims based on, inter alia, the district court's determination that the plaintiffs' statistical evidence "did not support a finding of companywide gender discrimination").
In Rodriguez v. National City Bank, 726 F.3d 372 (3d Cir. 2013), the Third Circuit explained why statistical correlation falls short where the "policy" under evaluation is one that permits localized discretion. There, the circuit court affirmed a district court's rejection of a class action settlement in a discriminatory lending case on the basis that the proposed settlement class could not be certified. Relying on Wal-Mart, the Rodriguez court rejected the plaintiffs' argument that their regression analysis sufficed to establish common causation because it controlled for all of the objective factors that could have influenced the cost of mortgage loans. Even assuming that truth of that assertion (which is the same claim the plaintiffs make here), the Third Circuit held that the regression analysis could not establish a common cause for the higher average loan costs charged to minorities because it could not account for subjective factors that may have influenced the pricing decisions made by the defendants' loan officers:
Even if Plaintiffs had succeeded in controlling for every objective credit-related variable... the regression analyses do not even purport to control for individual, subjective consideration. A loan officer may have set an individual borrower's interest rate and fees based on any number of non-discriminatory reasons, such as whether the mortgage loans were intended to benefit other family members who were not borrowers, whether borrowers misrepresented their income or assets, whether borrowers were seeking or had previously been given favorable loan-to-value terms not warranted by their credit status, whether the loans were part of a beneficial debt consolidation, or even concerns the loan officer may have had at the time for the financial institution irrespective of the borrower.
726 F.3d at 384.
The same holds true here. A policy of discretion affords brokers the opportunity to exercise their pricing discretion in a multitude of ways that are purely subjective. Perhaps the broker deems one borrower to be a better source of future referrals and so extends a slightly better deal. Another borrower may be a better negotiator, having obtained pre-loan counseling that is recommended by the FDIC.Another borrower may have options available, such as selling his home, that make him more price sensitive. Further, as Rodriguez points out, 726 F.3d at 384, factors that have nothing at all to do with the borrower may influence the exercise of the broker's pricing discretion. One broker's economic situation may dictate an approach to pricing that is more, or less, aggressive; another's pricing calculus may factor in the intrinsic rewards of helping people stay in their homes while others may be in it only for the money. The potential distinctions in how individual brokers exercise their pricing discretion are endless, which explains why a regression analysis cannot control for them. These sorts of factors cannot be ruled out as a factor in a pricing decision that is the product of individual discretion, so even if statistics prove that purely objective factors do not account for a disparate impact, it cannot be said that all such pricing decisions are the product of a common exercise of discretion by the brokers.
The plaintiffs do not purport to be able to control for subjective factors; rather, they contend that they have accounted, and can control, for all relevant objective factors and that there are no subjective factors to consider because this case is involves HECM's (Home Equity Conversion Mortgages), a species of reverse mortgages. Because HECM's are secured by a portion of a borrower's equity in the home, various credit factors that influence the pricing of conventional mortgage loans are not considered in pricing HECM's. The plaintiffs maintain that the limited number of objective factors that influence the pricing of HECM's makes it possible to construct a regression analysis that controls for all of them. But even assuming, arguendo, that a regression analysis controls for all objective factors that are material to a pricing decision, it cannot control for subjective factors because, by definition, information about those factors lies outside the objective loan data on which a regression analysis would be based. The plaintiffs' effort to distinguish Rodriguez on this basis plainly fails; in Rodriguez, the Court assumed that the statistical analyses that had been conducted controlled for all of the objective factors relevant to the loan pricing decision. 726 F.3d at 384. Notwithstanding that assumption, however, the Court recognized that a policy of discretion necessarily introduces subjectivity into the pricing equation. Plaintiffs' contention that subjective factors play no role in discretionary loan pricing decisions ignores the reality of human interaction as well as the multiple personal, environmental, and market factors that might affect a negotiation over the terms of a loan.
Plaintiffs' complaint that they are being required to prove the merits of their claims while prohibiting them from conducting the discovery necessary to do so thus misses the point: the discovery they seek will not permit them to prove common causation.Even if they were permitted to conduct the statistical discovery they seek, they would necessarily fall short of establishing causation without other evidence to show that the defendant's brokers exercised their discretion in a uniform manner. And the plaintiffs concede that there is no such evidence: "There simply is no other means in existence for discovering and showing whether Generation's brokers exercised their pricing discretion in a common manner." Dkt. 89 at 6-7.
The plaintiffs' assertion that the Court has "impose[d] on plaintiffs a requirement to show disparate treatment in connection with their disparate impact claim, " Motion at 4, is also off-base. The court's opinion was based on the plaintiffs' confessed inability to establish causation, not intent. That there is no requirement to prove discriminatory intent in a disparate impact case does not relieve the plaintiffs from the obligation to establish a common cause of the disproportionate impact. Wal-Mart and Bolden establish that, where the impact is the product of the exercise of independent discretion by many different actors, statistical evidence does not suffice to establish a common cause for the alleged discriminatory effect.
The plaintiff also reassert a point that the Court did not expressly address in its original opinion, namely, that discovery of statistical evidence of a disparate impact is required even to prove an individual disparate impact claim. The plaintiffs maintain that, even if this case cannot go forward as a class action, they are entitled to the same discovery in order to prove their individual claims and so there is no basis to bar access to the statistical data they seek. This point might have some resonance if there were a realistic possibility that the plaintiffs were interested in pursuing individual claims, but in this case it actually supports the Court's view that bifurcation is appropriate. In advancing this argument, the plaintiffs have forgotten (or simply have ignored) their averment that "individual cases are not economically feasible" to pursue. Third Amended Complaint, Dkt. 107, at ¶ 194. At a minimum, that concession casts substantial doubt on the plaintiffs' interest in pursuing their individual claims and warrants bifurcation in order to establish first and definitively whether a class claim may go forward. See Rule 23(c)(1)(A) (courts should make class certification decisions "at an early practicable time"). It would not be fair or just to bootstrap from an insubstantial prospect that individual claims will be pursued the enormous discovery burden that the plaintiffs seek to impose in this case. See, e.g., Harris v. Comscore, Inc., 2012 WL 686709, at *4 (N.D. Ill. Mar. 2, 2012) (unlikelihood that plaintiffs would pursue individual claims weighs in favor of requiring class discovery to precede merits discovery); Manual for Complex Litigation § 21.14 at 256 ("in cases that are unlikely to continue if not certified, discovery into aspects of the merits unrelated to certification... can create extraordinary and unnecessary expense and burden"). In view of the plaintiffs' concession about the economic viability of their individual claims, there seems little reason to indulge the fiction that the plaintiffs intend to pursue those individual claims absent a class claim. If, however, the putative class claim is put to rest (whether by withdrawal or dismissal), and the plaintiffs nevertheless opt to go forward with their individual claims, they will not be precluded by this order from seeking statistical evidence relevant to establishing disparate impact (scaled appropriately to the context of each individual's claim). The Court's bifurcation ruling does not prejudice the plaintiffs' individual claims in any way.
Finally, the Plaintiffs' contention that "the Court's dim' view of plaintiffs' ability to certify a class... appears to have led to its decision to bifurcate discovery" puts the conclusion before the predicate. The Court's reference to the plaintiffs' seemingly dim prospects for certifying a class is the result, not the cause, of its determination that the statistical analysis the plaintiffs sought "will not-even if it shows, as plaintiffs expect, that African-Americans and single women paid more for their loans with Generation than did others-suffice to glue' their claims together in in a way that satisfies Rule 23(a)(2)'s commonality requirement." It is the product as well of the plaintiffs' repeated concessions that they can propose no other means to establish a common cause of class injury. Where there is no reason to believe that "discovery could make a difference to the issue of certification in this case, [there is] no reason to delay the inevitable." Ladik v. Wal-Mart Stores, Inc., 291 F.R.D. 263, 273 (W.D. Wis. 2013) (granting defendant's motion to dismiss class certification claim on the pleadings because under Wal-Mart they could not establish a common cause for a disparate impact). Law suits without "a reasonably founded hope that the discovery process will reveal relevant evidence" should not go forward. Dura Pharmaceuticals, Inc. v. Broudo, 544 U.S. 336, 347 (2005) (internal punctuation and citation omitted). If, as the plaintiffs maintain, there is no evidence-other than statistical evidence of a disparate impact-to show that Generation's brokers exercised their pricing discretion in a common manner, that is a reason to stop the plaintiffs at the gates, not to permit them to inflict enormous costs by engaging in discovery that is bound to prove futile.
The plaintiffs suggest that the Court has concluded that "there is no way-either statistically, mathematically or empirically-to prove causation in support of a disparate impact claim." Not so. The Court has merely concluded-as Wal-Mart holds-that statistical analysis, no matter how rigorous, cannot establish that a company policy that permits brokers to set loan rates on the basis of individual, subjective, criteria, constitutes a common cause of injury across a class comprising customers of many different brokers. The Court therefore agrees with the defendant that discovery of the class-wide statistical data that the plaintiffs seek should not be permitted and class and merits discovery should otherwise be bifurcated. The motion for reconsideration is therefore denied. This ruling does not foreclose the plaintiffs from pursuing other evidence relevant to establish any of the requirements of Rule 23.