County Of Cook v. Bank of America Corporation

 IN THE UNITED STATES DISTRICT COURT FOR THE NORTHERN DISTRICT OF ILLINOIS EASTERN DIVISION COUNTY OF COOK, ) ) Plaintiff, ) ) v. ) No. 14 C 2280 ) BANK OF AMERICA ) CORPORATION, et al. ) ) Defendants. ) MEMORANDUM OPINION AND ORDER Nearly eight years ago, Cook County filed this action under the Fair Housing Act of 1968 (“FHA”) to recover for economic and non-economic injuries that it claims to have suffered as a result of defendants’ predatory and discriminatory scheme to strip equity from the homes of African American and Hispanic borrowers in Cook County (“minority borrowers”) and to foreclose disproportionately on their homes. The County alleged that defendants identified and targeted minority borrowers “using advanced data mining techniques and predictive analysis methodologies” and engaged in practices that encouraged: (a) unchecked or improper credit approval decisions for minority borrowers, which allowed minority borrowers to receive home loans they could not afford; (b) discretionary application of surcharges on minority borrowers of additional points, fees, and other credit and servicing costs over and above an otherwise objective risk-based financing rate for such loan products; (c) steering minority borrowers into higher cost loan products; and (d) undisclosed inflation of appraisal values of minority residences to support inflated loan amounts to minority borrowers. Compl. ECF 1 at ¶ 7. All of these practices allegedly increased the likelihood of default and foreclosure among minority borrowers. Id. See also id. at ¶ 103 (alleging additional discriminatory terms and conditions); ¶ 299 (same). As a result of the foregoing practices, the County claimed to have suffered injuries that included: (1) out-of-pocket costs to provide governmental services associated with foreclosed and/or vacant properties; (2) lost property tax revenue; (3) lost recording fee income; and (5) intangible injuries to the fabric of its communities. Id. at ¶ 408. After the County’s original complaint survived a motion to dismiss, see Cty. of Cook v. Bank of Am. Corp., 181 F. Supp. 3d 513 (N.D. Ill. 2015), the Supreme Court’s decisions in Bank of Am. Corp. v. City of Miami, Fla., 137 S. Ct. 1296 (2017) (“City of Miami”), and Texas Department of Housing & Community Affairs v. Inclusive Communities Project, Inc., 135 S. Ct. 2507 (2015) (“Inclusive Communities”), altered the legal landscape of the County’s claims. City of Miami confirmed the County’s Article III standing to pursue FHA claims for economic injuries but held that the statute’s proximate causation requirement limited the scope of its actionable injuries to those the County could prove were the direct result of the discriminatory conduct alleged. 137 S. Ct. at 1305-06. Inclusive Communities confirmed that the County’s disparate impact theory of discrimination is cognizable under the FHA but held that claims based on that theory are subject to a “robust causality requirement” that requires the County to do more than offer statistical evidence of racial imbalance; the County must also point to a specific policy and show that it created “artificial, arbitrary, and unnecessary barriers” to equality. 576 U.S. at 543. In the wake of these developments in the Court’s FHA jurisprudence, the County filed a Second Amended Complaint (“SAC”) in July of 2017. Count I of the SAC asserts the theory that defendants carried out an equity stripping scheme designed to enrich themselves at the expense of minority borrowers, who disproportionately suffered default and foreclosure, in violation of the FHA.1 In Count II, the County specifically challenges defendants’ facially neutral servicing and foreclosure practices, which the County likewise claims had a disparate impact on minority borrowers in violation of the FHA. 1 Although the County alleges that defendants’ equity stripping scheme was intentional, Count I is based on “statistical disparities in foreclosure rates resulting from Defendants’ equity stripping practices,” not on evidence of discriminatory intent. See Pl.’s Opp. ECF 622 at 3 n. 4 (ECF 622) And in Count III, the County claims that defendants’ equity stripping scheme amounts to intentional discrimination in violation of the FHA because it targeted minority borrowers for higher cost, higher risk loans based on their race and/or color. See SAC, ECF 177 at ¶¶ 403-551. Defendants moved to dismiss the SAC in its entirety. Although I rejected defendants’ arguments that the County failed to articulate either a plausible claim for any injury proximately caused by defendants’ alleged discrimination or the “robust” causality required to proceed on a disparate impact theory, I held that the majority of the injuries the County claimed—including the alleged erosion of its tax digest and costs associated with the provision of downstream social services—were “too remote in time, and too contingent on later events, to satisfy the “first step” directness requirement of City of Miami. Accordingly, I concluded that only a “narrow category” of the County’s alleged damages, namely, “the out-of- pocket costs it claims to have incurred in processing the discriminatory foreclosures,” had “a sufficient temporal and practical connection” to the challenged foreclosures to satisfy City of Miami’s proximate causation standard. I specifically identified “additional funding for the Cook County Sheriff to serve foreclosure notices and for the Circuit Court of Cook County to process the deluge of foreclosures” as examples of such costs, though I expressed skepticism that these losses were worth the cost of pursuing the County’s far-reaching claims. Mem. Op. and Order of 03/30/2018, ECF 204 at 19-20. I later clarified, pursuant to a motion by the County, that the County could also seek “out-of-pocket costs in serving eviction notices, conducting judicial and administrative foreclosure proceedings, and registering and inspecting foreclosed properties.” Order of 08/17/18, ECF 228 at 1. These decisions, read together, identified a closed set of “out-of-pocket” costs that I concluded the County could attempt to show—using the statistical regression analysis it heralded as capable of isolating the effects of defendants’ conduct from the influence of numerous other factors—were directly caused by defendants’ alleged discrimination. Discovery followed, punctuated by substantial motion practice, occasionally before me and frequently before the two magistrate judges successively assigned to this case. Defendants then filed a motion for summary judgment, which is now ripe for decision. Also pending are the parties’ respective Daubert motions, which I have considered in conjunction with their summary judgment arguments. For the following reasons, I grant defendants’ summary judgment motion and resolve the Daubert motions as set forth below. I. The overarching theme of the SAC is that defendants’ equity stripping scheme and discriminatory servicing and foreclosure practices caused skyrocketing foreclosure rates in Cook County from 2004 to approximately 2008, which disproportionately affected FHA-protected minority borrowers and communities with a high concentration of minorities. This avalanche of foreclosures, in turn, allegedly caused the County to suffer an increase in foreclosure-related expenses that it claims as damages resulting from defendants’ discriminatory lending practices. Defendants attack this theory on nearly every front. Their primary arguments are: 1) that the County has no evidence that any unified “equity stripping scheme” existed, much less one that targeted minority borrowers, nor any evidence of intentional discrimination, i.e., disparate treatment; 2) that even assuming Countrywide and/or Bank of America2 engaged in the 2 Throughout this opinion, “Countrywide” refers collectively to Countrywide Home Loans, Inc., Countrywide Bank, FSB, and/or Countrywide Warehouse Lending, while “Bank of America” refers to Bank of America, N.A., and/or BAC Home Loans Servicing, LP. Although the SAC also names as defendants Bank of America Corp., Countrywide Financial Corp., and Merrill Lynch & Co. (to which defendants refer collectively as the “Holding Companies”), and Merrill Lynch Mortgage Capital, Inc., and Merrill Lynch Mortgage Lending Inc. (the “Merrill Defendants”), the County does not dispute that it offers no evidence that these entities originated, serviced, or foreclosed on any of the at-issue loans. And while the County insists that these entities can practices the County identifies as elements of the putative scheme, the evidence is insufficient to allow a reasonable jury to conclude that those practices proximately caused minority borrowers to suffer foreclosure more frequently than similarly situated white borrowers; and 3) that the County has not proven any of the losses for which I held it could seek damages because a) it offers no evidence of any increase in its out-of-pocket expenses as a result of defendants’ allegedly discriminatory foreclosures, and b) its resource-shifting theory of loss is both legally flawed and factually unsupported. The County responds that the record developed in discovery, and, in particular, the opinions and statistical evidence proffered by its experts Dr. Gary Lacefield and Dr. Charles Cowan—both of whose opinions defendants seek to exclude under Fed. R. Evid. 702 and Daubert v. Merrell Dow Pharms., 509 U.S. 579 (1993)—establish that defendants’ equity stripping scheme caused minority borrowers to pay higher monthly payments on higher loan balances than similarly situated white borrowers, and that as a result, minority borrowers suffered delinquencies and foreclosures at higher rates. In the County’s view, the nevertheless be held liable based on defendants’ corporate relationships and/or the business dealings among them, its argument in this connection is wholly conclusory. Accordingly, summary judgment is appropriate as to these entities, and unless otherwise noted, my use of the term “defendants” in this opinion refers to Countrywide and Bank of America as defined above. record also shows that minority borrowers received fewer loan modifications and work-outs than white borrowers; were denied loan modifications under the Home Affordable Modification Program (“HAMP”) for which they were eligible; and were placed in foreclosure at higher rates than white borrowers. The County also relies on Dr. Cowan’s testimony to support its claim for damages to recover for the “resource shifting” that it claims was required to process the foreclosures resulting from defendants’ discriminatory practices. II. Summary judgment obviates the need for a trial when there “is no genuine dispute as to any material fact and the movant is entitled to judgment as a matter of law.” Fed. R. Civ. P. 56(a). See also Celotex Corp. v. Catrett, 477 U.S. 317, 322 (1986). A “material” fact is one that “might affect the outcome of the suit under the governing law.” Anderson v. Liberty Lobby, Inc., 477 U.S. 242, 248 (1986). A “genuine” dispute exists if there is sufficient evidence to allow a reasonable factfinder to decide in favor of the nonmoving party. Id. “Federal Rule of Civil Procedure 56 imposes an initial burden of production on the party moving for summary judgment to inform the district court why a trial is not necessary.” Modrowski v. Pigatto, 712 F.3d 1166, 1168 (7th Cir. 2013). Where, as here, the non-movant bears the underlying burden of persuasion, the movant need not “support its motion with affidavits or other similar materials negating the opponent’s claim.” Id. (quoting Celotex, 477 U.S. 323 (emphasis in original)). That is, the movant can discharge its initial burden “by ‘showing’—that is, pointing out to the district court—that there is an absence of evidence to support the nonmoving party’s case.” Green v. Whiteco Indus., Inc., 17 F.3d 199, 201 (7th Cir. 1994) (quoting Celotex 477 U.S. at 325)). Upon such a showing, the nonmovant must then point to evidence “sufficient to establish the existence of an element essential to that party’s case” to withstand summary judgment. Modrowski, 712 F.3d at 1168 (citation omitted). A. Integrated Equity Stripping Scheme Defendants’ lead argument is that the County lacks any evidence to support the cornerstone of its equity stripping claims in Counts I and III: the existence of an integrated scheme to strip equity from minority borrowers. At best, defendants submit, plaintiffs point to a concatenation of origination, underwriting, servicing, and foreclosure practices that Countrywide and/or Bank of America implemented at various times over the years (indeed, decades) during which they issued or purchased the challenged loans. This is not a trifling distinction. To the contrary, at the motion to dismiss stage, the County successfully withstood defendants’ argument that the SAC did not plausibly plead proximate causation by insisting that the SAC alleged “an integrated, discriminatory equity- stripping scheme that began with predatory lending and ended in the foreclosures that directly caused its injuries.” 03/30/2018 Mem. Op., ECF 204 at 8 (citing Pl.’s 09/14/2017 Resp., ECF 184 at 7). See also Cty. of Cook v. HSBC N. Am. Holdings Inc., 314 F. Supp. 3d 950, 961 (N.D. Ill. 2018) (allegations of a “comprehensive equity-stripping program” distinguished the County’s claims from those resting exclusively on lenders’ origination practices, as the latter “left open the possibility that the foreclosures...could have been caused by a wide array of factors outside of the lenders’ control.”).3 Indeed, the County criticized defendants for “misdirect[ing] the Court by breaking down and disjoining the 3 A third court in this district declined to dismiss similar claims at the pleading stage, holding that the County’s allegations targeting the defendants’ foreclosure practices—the subject of Count II of the SAC—were sufficient to plead proximate causation of its foreclosure-processing injuries. See Cty. of Cook, Illinois v. Wells Fargo & Co., 314 F. Supp. 3d 975, 984 (N.D. Ill. 2018) (allegations that Wells Fargo “exercised discretion over whether to grant loan modification requests to borrowers already behind on their payments and whether to foreclose on borrowers in default,” and thus “determined not only the number of homes in Cook County that would end up in default, but also the number that would end up in foreclosure” were sufficient to plead proximate causation, as this conduct necessarily “trigger[ed] certain obligations on the County’s part, including posting foreclosure and eviction notices, serving foreclosure summonses, executing evictions, and processing foreclosure suits.”) various components of the single equity stripping scheme, improperly treating each component as a distinct step in the causal chain of a foreclosure...and ignoring the core scheme allegation that loan defaults and foreclosures were the intended result” of the discriminatory scheme. Pl.’s Resp. ECF 184 at 1 (emphasis added). It was the existence of the integrated scheme, the County insisted, that created a “direct, single-link causal chain” between the foreclosures and the County’s alleged economic injury, i.e., its expenditure of “money in the form of foreclosure-related proceedings,” as well as the additional downstream losses the County claimed as damages. Id. at 7. See also id. at 15-16 (“The County sued Defendants for their discriminatory equity stripping scheme, and for their stand- alone discriminatory foreclosure practices, both of which are premised on the vacancy/foreclosure as the ultimate causal event to the County’s injuries.”). But nothing in the record the County has since developed suggests that defendants—who, it bears recalling, competed with each other in the loan business until Bank of America acquired Countrywide in 2008 and 2009—engaged in a coordinated scheme to provoke defaults and foreclosures by minority borrowers by locking them into loans they could not afford—a scheme that Dr. Lacefield admitted would “not make economic sense.” Lacefield Spolin Resp., ECF 573-15 at 15. No witness testified that such an integrated scheme existed. No document suggests that the practices the County challenges were interconnected elements of a unified effort to strip borrowers of the equity in their homes. And neither of the County’s liability experts—even assuming that their opinions are admissible—opines or offers any basis to conclude that the loan features or lending practices they deem discriminatory were part of unified equity stripping scheme. Indeed, the County does not dispute that it lacks affirmative evidence of the integrated scheme the SAC alleges. Nevertheless, it seeks to withstand summary judgment by miscasting defendants’ argument as resting on the absence of a written policy setting forth the equity stripping scheme. See Pl.’s Opp., ECF 622 at 7-8 (“Contrary to Defendants’ misconception, the County’s two equity stripping claims are not dependent on proof of a written policy permitting ‘equity stripping’ but rather are proven by evidence showing that Defendants’ practices in how they carried out their facially neutral policies resulted in a disparate and discriminatory impact in African American and Hispanic neighborhoods.’”) (emphasis in original). But the problem defendants identify is not that the County cannot point to a written policy. The problem is that the County’s theory of liability, and indeed, the proximate causation analysis set forth in City of Miami, require the County to prove a unified scheme that the record simply does not substantiate. The County tries to make up for this shortcoming by pointing in scattershot fashion to evidence that Countrywide and/or Bank of America engaged in a variety of practices over roughly a decade that adversely affected borrowers: marketing loan products to consumers who were not qualified for traditional loans; offering incentives to employees to increase loan volume among borrowers with lower credit scores; making or servicing high-risk loans; departing from underwriting standards; and/or failing to follow servicing guidelines. But even if a jury were persuaded that Countrywide and/or Bank of America engaged in one or more of these practices at some point between 2004 and 2012, nothing in the County’s submissions offers a basis for the jury to leap from such a finding to the conclusion that defendants carried out an integrated equity stripping scheme targeting minority borrowers. Having averted dismissal under City of Miami on the argument that the unified nature of defendants’ scheme is what allowed the County to establish a “direct, single-link causal chain” between defendants’ lending, servicing, and foreclosure practices on the one hand, and the County’s injuries on the other—and indeed, having disclaimed any theory of liability premised on the “disjoin[ted]” components of the scheme—the County cannot now stand exclusively on evidence of those components as proof of the unified scheme. The bottom line is that the County has not presented sufficient evidence to establish the causal link its equity stripping claims require between the conduct it challenges and the injuries it claims. For this reason alone, summary judgment of Counts I and III is appropriate. B. Intentional Discrimination Defendants seek summary judgment of the equity stripping claim in Count III for the additional reason that the County’s evidence does not give rise to a reasonable inference of intentional discrimination. To survive summary judgment on its disparate treatment claim, the County must come forward with evidence of “intentional discrimination, provable via either direct or circumstantial evidence.” Cty. of Cook v. HSBC N. Am. Holdings Inc., 314 F. Supp. 3d 950, 966 (N.D. Ill. 2018) (citations omitted). “Direct evidence is that which can be interpreted as an acknowledgment of the defendant’s discriminatory intent,” East-Miller v. Lake Cty. Highway Dep’t, 421 F.3d 558, 563 (7th Cir. 2005) (quoting Kormoczy v. Sec'y, U.S. Dep’t of Hous. & Urb. Dev. on Behalf of Briggs, 53 F.3d 821, 824 (7th Cir. 1995), which is to say, “a ‘smoking gun’ of discriminatory intent,” Cavalieri-Conway v. L. Butterman & Assocs., 992 F. Supp. 995, 1003 (N.D. Ill. 1998). The County tacitly concedes that it lacks such “smoking gun” evidence as discriminatory statements by a policymaker or explicitly race- based policies, see Pl.’s Opp., ECF 622 at 8 (observing that plaintiffs “rarely have actual evidence of discrimination, such as a written policy or an admission of impermissible animus”), and it makes no effort to controvert defendants’ evidence that their lending, servicing, and foreclosure policies were based on race-neutral, credit-related criteria. To the contrary, Dr. Lacefield agreed with the assessment of defendants’ expert, Sharon Stedman, that “both Countrywide and Bank of America had fair lending compliance programs that were consistent with industry and government standards.” Stedman Rpt., ECF 573-51 at ¶ 49; Lacefield Stedman Resp., ECF 573-2 at p. 25. See also Stedman Rpt. at ¶¶ 52-59 (describing policies reflecting oversight); ¶¶ 60-65 (describing policies reflecting risk identification, management and monitoring); ¶¶ 66-69 (describing policies reflecting training); ¶¶ 72-78 (describing defendants’ controls to prevent steering); ¶¶ 80-83 (describing controls to address discretionary pricing). Indeed, Dr. Lacefield frankly admitted that “[d]efendants had all of the right manuals in place, guidance prepared, and training for staff.” Lacefield Stedman Resp., ECF 573-52 at p. 29.4 4 Dr. Lacefield opines that the “resulting data, foreclosure rates, and sworn statements from the people in charge” indicate To withstand summary judgment of its disparate treatment claim, the County points to snippets of evidence drawn from various sources and argues that these materials, taken together with statistical disparities its experts observe in foreclosure rates, raise an inference of intentional discrimination. This argument is flawed on numerous fronts. 1. Data mining The County first points to defendants’ alleged use of “data mining” and other “highly sophisticated techniques” to identify and target minority populations for the purpose of marketing home loans. Pl.’s Opp., ECF 622 at 8-11. Setting aside that much of the evidence the County cites in this connection does not relate to the marketing of home loans, see, e.g., 2014 Bank of America presentation, “Compliance Office Forum Fair Lending Hot Topics,” ECF 618-7 at 3 (“Race Estimation – a game changer for fair lending risk in non-mortgage credit”), and that none of the cited materials indicates that minority borrowers were targeted for specific (e.g., risky or high-cost) loan products, see, e.g., Countrywide presentation, “African American Retail Campaign,” ECF 617-7 at BANACC0000169489 (identifying campaign that these policies and controls were not followed, but his opinions in this connection are flawed for the reasons I address elsewhere. objectives),5 if there is any authority for the proposition that soliciting business from minority prospects, or marketing in neighborhoods with a high concentration of minority residents, amounts to intentional discrimination in violation of the FHA, the County has not cited it. Indeed, it is not difficult to imagine an FHA action premised on a lender’s failure to do these things while soliciting business from white borrowers and marketing in predominantly white neighborhoods.6 2. Statistical disparities in issuance of high-risk loans Next, the County argues that statistical evidence that minority borrowers “were placed in specific types of high-risk subprime/nonprime loan products at a higher rate than similarly situated white borrowers,” despite evidence that defendants knew “such loans placed those borrowers at a higher risk of 5 Although I cite only these examples, I have reviewed the remaining materials the County cites in its Corrected Local Rule 56.1 (B)(3) Statement of Additional Material Facts ¶¶ 5-9 and conclude that they likewise fail to raise a reasonable inference of race-based targeting for specific loan products. 6 As Dr. Cowan notes in his Response Report, ECF 593-3, the OCC’s Comptroller’s Handbook on Fair Lending defines “redlining” as “a form of illegal disparate treatment in which a bank provides unequal access to credit, or unequal terms of credit, because of the race, color, national origin, or other prohibited characteristic(s) of the residents of the area in which the credit seeker resides or will reside or in which the residential property to be mortgaged is located,” and includes as an example: “A bank omits or excludes such an area from efforts to market residential loans or solicit customers for residential credit.” See https://www.occ.treas.gov/publications-and- resources/publications/comptrollers-handbook/files/fair- lending/pub-ch-fair-lending.pdf (last accessed January 20, 2022). delinquency, default, and foreclosure,” raises an inference of intentional discrimination. Pl.’s Opp., ECF 622 at 11. A threshold problem with this argument is that it relies heavily on the flawed analyses of Drs. Lacefield and Cowan, which I address at greater length in a later section. For present purposes, it suffices to observe that: 1) Dr. Lacefield’s analysis wholly misunderstands what it means to compare “similarly situated” borrowers; and 2) the County admits that Dr. Cowan does not opine that any defendant engaged in intentional discrimination or that his statistical analyses support a claim of intentional discrimination. See Pl.’s Resp. to Def.’s L.R. 56.1 Stmt., ECF 623 at ¶ 44. At all events, statistical disparities of the kind the County points to are rarely sufficient to raise an inference of intentional discrimination. Alston v. City of Madison, 853 F.3d 901, 907 (7th Cir. 2017) (“disparate impact alone is almost always insufficient to prove discriminatory purpose.”) (citing Washington v. Davis, 426 U.S. 229, 239 (1976). Even assuming that minority borrowers disproportionately received loan products with features the County characterizes as risky, and assuming further that defendants knew recipients of such loans were more likely to enter default and foreclosure than recipients of traditional loan products, “[d]iscriminatory purpose means more than simple knowledge that a particular outcome is the likely consequence of an action; rather, discriminatory purpose requires a defendant to have selected a particular course of action at least in part because of ... its adverse effects upon an identifiable group.” Id. (internal quotation marks and citations omitted) (emphasis added). This is not an inference that reasonably emerges from the County’s statistical evidence. Even taken at face value, this evidence comes nowhere near establishing that “almost all minorities” were “negatively affected” by the practices the County challenges, while “almost no whites” were negatively affected. Id. at 908. See also Chicago Tchrs. Union v. Bd. of Educ. of the City of Chicago, 14 F.4th 650, 657-58 (7th Cir. 2021) (evidence that school layoffs disproportionately impacted African Americans and that Chicago’s Board of Education knew the layoffs would have a disparate impact on that group did not raise a triable issue of intentional discrimination). Moreover, the County “never explains how such knowledge, even if proven, would demonstrate that [defendants] intended to discriminate,” id. at 658, given the uncontroverted evidence that their lending, servicing, and foreclosure policies and practices were based on race-neutral criteria. 3. Financial incentives The County’s next argument—that intentional discrimination can be inferred from evidence of financial incentives that defendants provided loan originators to increase the volume of high-risk loans to minorities, see Pl.’s Opp. ECF 622 at 17-20— is wholly lacking in evidentiary support. Uncontroverted evidence reveals that loan originators’ compensation was based on the terms of the loans, as Dr. Lacefield himself explains, see Lacefield Rpt. at ¶ 23 (financial incentives “increased based upon the risk level of the mortgage product-the higher the risk, the greater the incentive”), and did not factor in the borrowers’ race or ethnicity. Indeed, the majority of the materials the County cites address policies reflecting differences in broker compensation as between nonprime versus prime loan origination and make no mention of race. See, e.g., Pl.’s Resp. to Def.’s L.R. 56.1 Stmt. ECF 623 at ¶ 70 (citing, inter alia, ECF 617-13 BANACC0000224250 at -224250-51 (reflecting higher maximum broker compensation rates for nonprime loans than for prime loans) and ECF 620-5, Miller Tr. 191:22-192:7 (confirming that brokers could earn 1% higher compensation for nonprime loans versus prime loans)). The County purports to show that within certain categories of loans, brokers actually received higher compensation rates for loans made to minorities than for loans made to non-minorities. Id. (citing, inter alia, ECF 618-11 and 618-13). But this ex post comparison does not evince intentional discrimination for substantially the reasons discussed above: disparate outcomes do not, per se, show intentional discrimination. Nor does the County’s remaining evidence add anything to support such an inference. For example, the County points to an email from Countrywide executive David Doyle, which includes text stating that “paying higher [broker compensation] on PayOption arms” posed a “significant risk, especially on refinances.” Pl.’s Opp. ECF 622 at 19 (quoting Doyle email of 05/31/2005, ECF 617-18). Although the County accurately quotes from Doyle’s email, a fuller reading of his message belies the inference the County suggests. The selected text appears in response to an email from another Countrywide employee, who writes: “I read the part about paying higher on PayOption Arms. I know the reason you would want to do that. We will want to make sure that it does not cause other issues such as stearing (sic) of borrowers.” Smith email of 05/25/2005, ECF 617-18. Doyle responds: “I agree with you that this is a significant risk, especially on refinances. We will be as thorough as we can be in our mitigation of this risk.” Doyle email of 05/31/2005, ECF 617-18 (emphasis added). Read as a whole, this exchange does not evoke intentional discrimination against minority borrowers. To the contrary, it suggests that Countrywide’s management was mindful of the potential for brokers to steer borrowers of any race toward products that yielded higher commissions, and that it took measures to mitigate that risk. 4. Departure from underwriting standards The County’s disparate treatment claim also gains no traction from evidence that defendants “loosened” their underwriting guidelines and increased the use of exceptions to obtain approval for high-risk loans that did not meet standard guidelines. See Pl.’s Opp. ECF 622 at 20-23. Even assuming a jury were to credit the County’s evidence and infer from it that defendants had a practice of extending credit to borrowers who did not have the ability to repay their mortgage loans, nothing in that evidence suggests that the practice targeted minorities in particular. 5. Failure to comply with HAMP guidelines and HUD requirements The County next submits that intentional discrimination can be gleaned from evidence that defendants failed to comply with HAMP guidelines and servicing requirements mandated by HUD. See Pl.’s Opp. ECF 622 at 23-24. This argument is equally meritless. If there is any evidence supporting the County’s claim that “race and ethnicity were factors considered by both Bank of America and Countrywide in determining borrowers’ eligibility for loan modifications and whether to grant or deny loan modification requests or place the loan in the foreclosure process,” the County has not cited it. Instead, the County once again relies heavily on statistics drawn from Dr. Lacefield’s analysis of loan outcomes and concludes that minority borrowers were treated differently from white borrowers. But Dr. Lacefield’s analysis in this connection (as in others) is replete with assumptions he does not test, for which he offers no evidence, and that in some instances are belied by his own data. I address these flaws in further detail in a later section, but offer an example here to illustrate why his opinion offers no basis for inferring intentional discrimination. Explaining his use of the “delimiter” SD-3 to evaluate whether defendants’ servicing practices were discriminatory, Dr. Lacefield states, “studies have indicated that the ‘failure to submit completed applications’ was the most prevalent reason modifications were declined. If ‘failure to submit completed applications’ were higher for minority populations[,] then those modification applications may be predatory/discriminatory because that could mean that minority applicants were not worked with as vigorously to resolve those issues as were white applications.” Lacefield Rpt. ECF 560-1 at ¶ 138 (emphasis added). But this observation does not raise an inference of discrimination because Dr. Lacefield offers no facts to substantiate his speculation that defendants might not have worked with minority borrowers as vigorously as they did white borrowers. See Gopalratnam v. Hewlett-Packard Co., 877 F.3d 771, 788 (7th Cir. 2017) (expert opinion that “several manufacturing processes ‘can cause’ an internal short circuit,” was “simply too speculative” to support his opinion that ‘such must have occurred here’”) (original emphasis)). Perhaps the reason Dr. Lacefield does not elaborate on this point is that the data do not support his premise. To the contrary, certain of Dr. Lacefield’s data undercut the inference that applications filed by minority borrowers were denied for “failure to submit completed applications” at higher rates than applications by white borrowers. See Appendix 5 to Lacefield Rpt., Tbl. SD-3, ECF 560-8 at PageID #10997 (reflecting that in census tracts with between 51-70% minority concentration, 20.6% of applications by white borrowers were rejected for failure to submit completed applications, while only 14.1% of applications by African American borrowers and 20.5% of applications by Hispanic borrowers were rejected for this reason). Dr. Lacefield ignores these data and cherry-pick others to opine that his application of delimiter SD-3 yields statistically significant race-based disparities in modification outcomes. ECF 560-1 at ¶ 150. See also Def.’s L.R. 56.1 Stmt. at ¶ 42 (citing undisputed data from Tbl. SD-1, ECF 560-8 at PageID #10994 revealing statistically lower rates of modification approval for white borrowers than for minority borrowers, and from Tbl. SD-6, ECF 577-1 at PageID #16471, showing that the proportion of loans identified as seriously delinquent and foreclosed was higher for white borrowers than for minority borrowers).7 Having examined both Dr. Lacefield’s data and the County’s remaining evidence concerning defendants’ servicing practices, I conclude that they do not controvert defendants’ factual statements that “[n]either race nor ethnicity was a factor used by [Bank of America/Countrywide] in evaluating borrowers for loan modifications.” Def.’s L.R. 56.1 Stmt., ECF 577 at ¶¶ 17-18 (citing Buchanan Tr., ECF 573-39 at 184:14-21 (“[R]ace was not a factor in evaluating a borrower for modification. It had nothing to do with it. . . . Not [the borrower’s] location, nothing.”); Haumesser Tr., ECF 573-16 at 76:14-77:5 (race was “not part of the decision process” for loan modifications) and 99:8-21 (Bank of America’s loan modification systems did not list the borrower’s race or ethnicity); Guidici Tr. 573-40 33:21-34:5 (“we didn’t have access to any of the information that . . . would allow us to know whether [the borrower] was a minority account or not.”). In addition to the evidentiary shortcomings discussed above, the County’s argument in connection with its disparate 7 These data illustrate one of several flaws in Dr. Lacefield’s analysis that I do not address in my Daubert discussion below, which is that he cherry-picks data that support his conclusions while ignoring those that do not. See Courchane Rpt. ECF 560-11 at Table A2.3 Dr. Lacefield’s Servicing Delimiters (“Dr. Lacefield looks at particular segments of census tract minority percentages, ignoring all the ones with no differences”). treatment claim is bereft of any meaningful legal analysis. The County’s only case citations appear in a section captioned, “A Reasonable Jury Could Find Discriminatory Intent from the Statistical Disparities in Defendants’ Foreclosure Rates,” where it cites E.E.O.C. v. O & G Spring and Wire Forms Specialty Co., 38 F.3d 872, 876 (7th Cir. 1994), and Chicago Tchrs. Union, Loc. 1, Am. Fed’n of Tchrs., AFL-CIO v. Bd. of Educ. of City of Chicago, No. 12 C 10311, 2021 WL 1020991, at *17 (N.D. Ill. Mar. 17, 2021), for the proposition that statistical imbalances so stark as to be “unexplainable on grounds other than race” raise an inference of intentional discrimination. But neither of these cases supports the County’s conclusory argument. The County does not discuss, quantify, or even cite to any of the statistical disparities its experts identify to illustrate their supposed “starkness,” nor does that inference emerges naturally from the data as a whole. Indeed, the County admits that “[c]ertain of Dr. Lacefield’s servicing delimiters are present at higher rates for loans to White borrowers than loans to African American or Hispanic/Latino borrowers,” as reflected, for example, in Table SD-1, which shows that in the aggregate, a lower percentage of white borrowers received loan modifications than minority borrowers. See ECF 560-8 at PageID #10994. In view of data such as these, the County’s unadorned statement that “the evidence of statistically significant disparities is stark enough for a reasonable jury to find Defendants’ discriminatory intent,” simply does not warrant the inference it seeks. For at least the foregoing reasons, summary judgment of Count III of the SAC is appropriate. C. Disparate Impact The County’s disparate impact claims assert that defendants’ facially neutral lending policies and practices caused statistical disparities between minority and white borrowers in violation of the FHA. Count I is based on defendants’ putative integrated equity stripping scheme,8 while Count II is based exclusively on defendants’ servicing and foreclosure policies. To withstand summary judgment on either claim, the County must offer evidence to show that the challenged practices have a “‘disproportionately adverse effect on minorities’ and are otherwise unjustified by a legitimate rationale.” Texas Dep’t of Hous. & Cmty. Affs. v. Inclusive Communities Project, Inc., 576 U.S. 519, 524 (2015) (quoting Ricci v. DeStefano, 557 U.S. 557, 577 (2009)). Additionally, the County must show a “robust” causal connection between defendants’ practices and the disparate impact. Id. at 542. “A plaintiff who fails to...produce statistical evidence 8 As explained above in Section II. A., this claim fails at the threshold for the County’s failure to present evidence sufficient to enable a jury to conclude that such a scheme existed. In this section, I address additional flaws in the County’s disparate impact claims. demonstrating a causal connection cannot make out a prima facie case of disparate impact.” Id. at 543. The County seeks to prove its disparate impact claims using the statistical analyses of Drs. Lacefield and Cowan, which defendants ask me to exclude pursuant to Fed. R. Evid. 702 and Daubert v. Merrell Dow Pharms., 509 U.S. 579 (1993). Accordingly, I begin with the admissibility of these experts’ opinions. In Daubert, the Supreme Court held that the Federal Rules of Evidence “assign to the trial judge the task of ensuring that an expert’s testimony both rests on a reliable foundation and is relevant to the task at hand.” Id. at 597. Elaborating on the Daubert framework in Kumho Tire Co. v. Carmichael, 526 U.S. 137 (1999), the Court explained that to satisfy the “twin requirements” of relevance and reliability, the expert must “‘employ[ ] in the courtroom the same level of intellectual rigor that characterizes the practice of an expert in the relevant field.’” Adams v. Ameritech Servs., Inc., 231 F.3d 414, 423 (7th Cir. 2000) (quoting Kumho Tire, 526 U.S. at 152). To discharge its gatekeeping function, a district court “must examine (among other things) the expert’s qualifications, the methodologies she used, and the relevance of the final results to the questions before the jury.” Id. The proponent of the expert testimony bears the burden of showing, by a preponderance of the evidence, that the Daubert standard is met. Lewis v. CITGO Petroleum Corp., 561 F.3d 698, 705 (7th Cir. 2009). 1. Dr. Lacefield The County proffers Dr. Lacefield as “a highly qualified fair housing and fair lending expert” with “extensive experience, including as Senior Civil Rights Analyst and Supervisor of Lending Investigations at the U.S. Department of Housing and Urban Development (“HUD”), in investigating, and training federal regulators and law enforcement on how to investigate, Fair Housing Act violations by mortgage lenders[.]” Pl.’s Resp., ECF 587 at 1. Without objection or contradiction from the County, defendants summarize the methodology Dr. Lacefield employed in this case as involving: (i) looking at the loan data for a series of lending and servicing characteristics, which he calls “delimiters,” that he identifies as hallmarks of “predatory” loans or servicing outcomes, then (ii) comparing the relative presence of those “delimiters” in the census-tract population of all White borrowers, on the one hand, to all African American and Hispanic borrowers, on the other hand, and (iii) designating every minority loan with a “delimiter” in that neighborhood as an instance of lending or servicing discrimination. Def.’s Mot., ECF 559 at 2. Defendants assail the reliability of this methodology on multiple grounds. Their lead argument is that Dr. Lacefield’s use of “delimiters” to conduct a fair lending analysis is unprecedented, unpublished, and unrecognized by any other expert in the field. Compounding these flaws, defendants add, is that Dr. Lacefield purports to detect discriminatory patterns using a bivariate statistical model that analyzes the presence of his “delimiters” without controlling for other factors, furthering diminishing the reliability of his conclusions. In addition, defendants argue that Dr. Lacefield misunderstands what it means to compare “similarly situated” borrowers, rendering the conclusions he draws from statistical disparities both unreliable and legally irrelevant. “When evaluating the reliability of expert testimony, the district court must make a preliminary assessment of whether the reasoning or methodology underlying the testimony is scientifically valid.” Kirk v. Clark Equip. Co., 991 F.3d 865, 873 (7th Cir. 2021). Daubert set forth “a non-exhaustive list of guideposts to consult in assessing the reliability of expert testimony: (1) whether the scientific theory can be or has been tested; (2) whether the theory has been subjected to peer review and publication; and (3) whether the theory has been generally accepted in the relevant scientific, technical, or professional community. Am. Honda Motor Co. v. Allen, 600 F.3d 813, 817 (7th Cir. 2010) (citing Daubert, 509 U.S. at 593–94). Dr. Lacefield’s methodology fails to satisfy any of these criteria, raising serious doubts about its reliability. In his expert report, Dr. Lacefield describes his “delimiters” as “red flags” that were “developed from the lending audit criteria used by the U.S. Department of Housing and Urban Development (HUD) to review the underwriting standards” of all lenders, and that were “designed to identify HUD’s highest risk scale consisting of four risk levels of loan audits.” ECF 560-1. at ¶¶ 82-83. Dr. Lacefield testified that he formulated the delimiters based on “a written list of red flags” found in an investigative manual HUD used at the time he worked there (from 1991-1999, see Lacefield curriculum vitae, ECF 560-2 at 11), which he said “any investigator conducting a fair lending investigation would have to look at in order to determine whatever – whatever the issue was they were examining.” Lacefield Tr., ECF 560-19 at 204:20-205:4. But if this “written list” or the decades-old HUD manual Dr. Lacefield describes is anywhere in the record or otherwise to be found, the County has not pointed me to it.9 In the end, the County identifies no written materials that explain or apply Dr. Lacefield’s delimiter-based analysis. That the delimiters were derived from “red flags” described in a manual Dr. Lacefield consulted in the 1990s does not establish that the specific 9 Defendants state that the County has produced no such document, nor have defendants identified any HUD manual describing Dr. Lacefield’s bivariate delimiter methodology. Lacefield Daubert Mot., ECF 603 at 4-5. methodology he used is generally accepted in the field of fair lending examination. And indeed, Dr. Lacefield acknowledged that he is unaware of any regulators, industry participants, or academics who have employed his methodology to conduct a fair lending analysis such as his any time in the past thirty years. Nor could Dr. Lacefield identify a single peer-reviewed academic paper that supports, recommends, or discusses his methodology. See generally Lacefield Tr., ECF 560-19. at 130-136; 202-204. None of the other experts in this case—including the County’s other liability expert, Dr. Cowan—was familiar with Dr. Lacefield’s delimiter-based methodology. See, e.g., Courchane Rpt. ECF 560- 11 at ¶ 28 (Dr. Lacefield’s “approach is unlike any other that I have seen used to assess fair lending issues, either by a government agency, a peer-reviewed academic paper, or an industry participant.”); Stedman Rpt., ECF 560-12 at ¶¶ 17, 33, 38 (same); Cowan Tr., ECF 560-20 at 227:20-228:21 (testifying that he was not familiar with the term “delimiter” in the context of fair lending analysis, and that to his knowledge, no federal regulator, academic, or court had ever used “delimiters” in that context). In short, there is no indication in the record—other than Dr. Lacefield’s own say-so—that Dr. Lacefield’s delimiter-based methodology has been used by anyone other than Dr. Lacefield himself. That is not a hallmark of reliability. See Allen, 600 F.3d at 818 (7th Cir. 2010) (expressing “definite reservations” about the reliability of a methodology that the plaintiff’s expert developed and was the only one to use); Chapman v. Maytag Corp., 297 F.3d 682, 688 (7th Cir. 2002) (excluding expert opinions based on a theory that was “novel and unsupported by any article, text, study, scientific literature or scientific data produced by others in [the expert’s] field.”). My own review of the expert materials in this case confirms not only that Dr. Lacefield’s methodology is untested and unheard-of by others in his field, but also that it is substantively unsound. Among its most salient flaws is his use of a simplistic, bivariate model to test for discriminatory impact. Dr. Lacefield explained that he “took the entire data set” of loans and “subjected them to” his various delimiters. Id. at 166:1-2. He then “evaluated whether the prevalence of these delimiters was different for White and minority loans using statistical tests,” and if there was a statistically significant difference, he “considered loans to be discriminatory on the basis of [the] delimiters if the loan for a minority group was flagged with the delimiter more often than the White group[.]” Lacefield Rpt., ECF 560-1 at ¶ 114.10 But 10 Dr. Lacefield explains that because a significant portion of the loans in the data set did not contain information on the this analysis makes no attempt to control for numerous other variables, such as differences in borrowers’ debt-to-income ratios (DTI), loan to value ratios (LTV), or credit score, which even Dr. Lacefield recognized could account for the statistical disparities he observed in the data. For example, Dr. Lacefield admitted that differences in origination outcomes might be explained by differences in creditworthiness. Lacefield Tr., ECF 560-19 at 144:23-149:11 (identifying DTI, LTV, positive credit history and other potential nondiscriminatory explanations for disparities with respect to delimiter 1). See also Lacefield Courchane Resp., ECF 560-15 at 14 (“I agree with Dr. Courchane’s statement that many factors such as the cost of funds and sale of mortgage loans to the secondary market, credit risk, down payment levels, prepayment risk, and servicing costs can all borrower’s race, he used census tract data to “estimate” borrower race. That is, if the subject property was located in a census tract with greater than 50% minority concentration, Dr. Lacefield assumed that the loan was a minority loan for purposes of evaluating disparities in the presence of these delimiters. Lacefield Rpt. 560-1 at ¶ 117. As Dr. Lacefield acknowledged, this methodology meant that white borrowers could have “ended up on the list” of loans Dr. Lacefield created that “represented situation where [defendants] had discriminated against the borrowers.” Lacefield Tr. ECF 560-19 at 185:9-14. But Dr. Courchane opines without contradiction from Dr. Lacefield that “neighborhood demographics are not a valid or accepted way to proxy of the race of actual borrowers.” Courchane Rpt., ECF 560- 11 at ¶ 85. Indeed, as Dr. Cowan’s analysis reflects, in a “neighborhood category of majority minority (51%-70%), the largest proportion of borrowers are White (41%), while only 24% are either African-American or Hispanic,” illustrating why inferring race based on census tract information is an unreliable methodology. Id. impact mortgage loan prices.”).11 For that reason, all federal financial agencies rely on multivariate regression analysis to examine fair lending compliance. Courchane Rpt. ECF 560-11 at ¶ 60.12 Dr. Lacefield’s simplistic delimiter analysis also fails to account for either the macroeconomic or the individualized reasons that he concedes influence the likelihood of borrower default. See Courchane Rpt. ECF 560-11 at ¶¶ 17-18 (identifying several “macroeconomic events during the period of study [that] caused many borrowers to default on their loans” and observing that “it is well-recognized that the reasons individual borrowers default and are unable to pay their mortgages is very often the result of post-closing life events as they occur in 11 Dr. Lacefield goes on to state, “[h]owever my analysis focuses on the primary factors that cause borrowers to default on their mortgages.” Lacefield Courchane Resp., ECF 560-15 at 14. But this statement merely assumes Dr. Lacefield’s conclusion that the factors captured by his “delimiters” are the primary cause of borrower default. 12 Dr. Lacefield does not meaningfully opine otherwise. The County seizes on Dr. Lacefield’s testimony that “the interagency guideline on fair housing, which covers OCC, OTS, the FFIEC, FDIC, credit unions, it says you can use the multi-regression analysis or other statistical methodologies.”) Pl.’s Resp. ECF 587 at 6, n.4 (quoting Lacefield Tr. ECF 560-19 at 133:22-134:6) (plaintiff’s emphasis). But Dr. Lacefield’s reference to provisions of the Federal Financial Institutions Examination Council’s (FFIEC) Interagency Fair Lending Examination Procedures contemplating the use, “under an agency’s policy,” of “other statistical methodologies” that employ “the agency’s specialized procedures,” ECF 588-2 at PageID #16858, does not establish general acceptance of Dr. Lacefield’s bivariate delimiter methodology to evaluate fair lending compliance. the larger context of the economy”); Lacefield Courchane Resp., ECF 560-15 at 2 (agreeing that the macroeconomic factors Dr. Courchane cites “had an impact on the high default rate and foreclosures”). Dr. Lacefield claims that he “can tie most of those issues back to the appraisal, origination, and underwriting.” Id. But even setting aside that I have already foreclosed recovery for the County on the theory that defendants’ alleged discrimination was the prime mover that triggered the adverse macroeconomic events Dr. Courchane cites, Dr. Lacefield’s stream-of-consciousness narrative in response to Dr. Courchane’s opinions in this connection do not meaningfully rebut them.13 13 The text following Dr. Lacefield’s assertion that he can tie most macroeconomic issues back to defendants’ conduct illustrates the tenor of his response to Dr. Courchane’s opinions. It reads as follows: For example: 1. Borrowers lost home equity as nationwide house prices fell [prices fell for a few key reasons: over-valued collateral and the borrower placed into homes they never really had the ability to repay-or maintain led to multiple foreclosures- reducing the value of the home-not the debt owed] and they found it difficult to sell or refinance homes [couldn’t sell because of the foreclosures in the neighborhood, turning some properties into rentals- further depressing the market. People had to refinance because of the tens of thousands of adjustable rate mortgages Defendants ‘qualified’ borrowers for. Let’s say after three years you had to refinance, get new mortgage (home would value less than owed), or get a new mortgage if they wanted if they wanted (sic) to stay in the home][the problem with refinancing is the requirement to have 20% equity or money down. Most Nor is Dr. Lacefield’s bivariate delimiter methodology capable of accounting for the individualized reasons that some of the borrowers whose loans he identifies as discriminatory actually defaulted on those loans. Of the 87,311 loans Dr. Lacefield identifies as discriminatory, see Cox Decl., ECF 560 at ¶ 9, 33,465 contain borrower-reported information about the reasons for default. Almost half of these (48.8%) attribute the default to “reduction of income,” while another 11.7% reported “unemployment” as the reason for default. Courchane Rpt., ECF 560-11, Table 8. Dr. Lacefield’s simplistic methodology does not account for these factors or seek to show that the borrowers who attributed their default to these concrete issues (or others folks I know wouldn’t be able to come up with 1% of the property value much less 20%. See statement above for reasons market depressed.]. when they had loan balances in excess of the market value of the home. (sic) The house price declines caused many foreclosures,[most of these foreclosures were from two categories of borrowers: , (sic) first group of borrowers were having to refinance out of an ARM product and the second group of borrowers were placed in loans they never had the ability to repay and maintain] the (sic) as well as causing some borrowers to turn to short sales (selling homes for less than the outstanding loan balances) [because the homes were overvalued to start with and other foreclosures in the neighborhood]or strategic defaults (choosing not to pay as they owed more than the home was worth). In addition, rising unemployment rates across the country led to high levels of job loss or to moves that were required for job mobility.[Job loss created, in part, because a large number of these families had employment tied in some fashion with the housing and real estate environment]. noted, such as illness and marital difficulties) would have been any less likely to enter foreclosure if their loans or loan histories had had none of the features corresponding to Dr. Lacefield’s delimiters. Dr. Lacefield’s only response to the evidence of these individualized factors is his unsupported assumption that “over time, all of these life experiences will happen to all families regardless of their race and ethnicity, equally.” Lacefield Courchane Resp. ECF 560-19 at 2. But that assumption is “true only if no other factor relevant to [those life experiences] is correlated with [race],” Sheehan v. Daily Racing Form, Inc., 104 F.3d 940, 942 (7th Cir. 1997), and both of the County’s experts testified that such correlations do exist. Lacefield Tr., ECF 560-19 at 249:14-21 (acknowledging that “the impact of loss of employment is not going to fall equally on African-Americans and Hispanics,” because “across the board, and throughout history,” these populations suffer post-closing unemployment more frequently than whites); Cowan Tr., ECF 564-3 at 256:8-22 (acknowledging that some borrower “distress events” may affect minorities more frequently than non-minorities). See also Courchane Rpt., ECF 560-11 at ¶ 111 (loan data “shows that when one examines the same loan product across different racial/ethnic groups, minority borrowers defaulted at higher rates than non-Hispanic white borrowers who selected the same loan product. This means it was likely some factor other than lender conduct that caused minority borrowers to default at higher rates. Possible alternative factors include higher rates of unemployment, higher rates of income reduction, lower levels of liquid assets, and a host of other individualized factors.”) (emphasis in original). Finally, it is clear that Dr. Lacefield’s methodology does not compare “similarly situated” borrowers as courts construe that phrase when applying federal antidiscrimination laws. “To raise an inference of discrimination, statistical comparators must be directly comparable to the plaintiff in all material respects,” Purtue v. Wisconsin Dep’t of Corr., 963 F.3d 598, 603 (7th Cir. 2020) (internal quotation marks and citation omitted). See also Williams v. Bd. of Educ. of City of Chicago, 982 F.3d 495, 505 (7th Cir. 2020) (comparators must be “similar enough to eliminate confounding variables”) (citation omitted). In the context of fair lending, borrowers are similarly situated if they have “similar underwriting and borrower characteristics.” City of Oakland v. Wells Fargo & Co., 14 F.4th 1030, 1033 (9th Cir. 2021). At his deposition, Dr. Lacefield agreed that all peer- reviewed articles say that “to conduct a statistical analysis of lending discrimination,” the populations compared had to be similarly situated. Lacefield Tr., ECF 560-19 at 42:5-10. He stated that he respected this principle in his investigation by applying “one underwriting factor or a series of underwriting factors to the entire universe of...loans,” id. at 43:1-3, and opined that “when you apply the same attribute across the board, then you are treating everyone similarly and trying to determine whether or not based upon that factor...there is a discriminatory effect or impact.” Lacefield Tr., ECF 560-19 at 43:5-9. But testing all borrowers across the board is not the same as ensuring that the borrowers being tested are similarly situated. Dr. Lacefield may have done the former, but he did not do the latter, which is what the law requires. The County’s insistence that Dr. Lacefield accounted for borrower characteristics such as FICO score, LTV ratios, and DTI ratios is also misguided. See Resp., ECF 587 at 10-11. It is true that Dr. Lacefield defined several of his delimiters by those features. See Lacefield Rpt. at ¶¶ 103, 105-06. But he did not control for those features when applying any of his other delimiters or do anything to isolate and account for their impact on loan outcomes. This is a fatal methodological flaw in a study intended to examine the relationship between a lender’s race discrimination on the one hand and a borrower’s foreclosure on the other. See Sheehan, 104 F.3d at 942 (7th Cir. 1997) (excluding expert opinion that failed “to correct for any potential explanatory variables” other than race, but merely “equat[ed] a simple statistical correlation to a causal relation.”). In short, Dr. Lacefield’s statistical analysis does not reflect the “care that a statistician would use in his scientific work, outside of the context of litigation.” Id. The foregoing is not an exhaustive account of the methodological and analytical flaws in Dr. Lacefield’s report, but rather a representative illustration of its insuperable shortcomings. The problem is not, as the County contends, that defendants’ experts disagree with Dr. Lacefield’s conclusions. The problem is that Dr. Lacefield’s methodology has never been used, much less approved, by anyone else in the field of fair lending and is fundamentally incapable of generating statistical data that would assist the jury in deciding whether there exists a “robust causal connection” between defendants’ practices and the race-based disparate impact the County seeks to prove. Accordingly, defendants’ motion to exclude all of Dr. Lacefield’s opinions is granted. 2. Dr. Cowan’s Liability Opinions Dr. Cowan is an expert in statistical research and design whom the County engaged to examine whether defendants: “(i) initiated foreclosures on Minority borrowers at higher rates than on White borrowers; (ii) issued higher cost loans to Minority borrowers, and (iii) issued higher-risk products to Minority borrowers.” Cowan Rpt., ECF 564-1 at ¶¶ 1, 5. The loan population Dr. Cowan analyzed comprised approximately 365,000 loans “originated or purchased by Defendants.” Id. ¶ 13. He summarizes his opinions as follows:  Defendants foreclosed on Minority loans at higher rates than on White loans, after controlling for factors associated with the credit-risk profile of the borrower.  Similarly, foreclosure rates are higher in neighborhoods with greater concentration of minorities, controlling for those same factors.  Defendants charged higher Annual Percentage Rates (APRs) to minority borrowers when compared to similarly situated White borrowers.  In the same way, APRs are higher in neighborhoods with a greater concentration of minorities.  Minority borrowers were issued higher-risk products at higher rates than similarly situated White borrowers.  In like fashion, the proportion of higher-risk products is higher in neighborhoods with greater concentration of minorities. Id. at ¶ 3. Defendants raise a battery of arguments targeting the methodology Dr. Cowan used to arrive at these conclusions. Their most salient criticism is that Dr. Cowan analyzed loan data on an aggregated basis, which is to say, across multiple lenders, multiple products, and multiple decades—an approach that Dr. Cowan himself admits no agency or regulator uses to conduct fair lending analyses. See Cowan Dep., ECF 564-3 at 247:5-8; 249: 16- 23. Indeed, the FFIEC Interagency Fair Lending Examination Procedures, which establish the procedural framework for federal agencies to use when conducting fair lending examinations, provide: Examiners should tailor their sample and subsequent analysis to the specific factors that the institution considers when determining its pricing, terms, and conditions. For example, while decisions on pricing, and other terms and conditions are part of an institution’s underwriting process, general underwriting criteria should not be used in the analysis if they are not relevant to the term or condition to be reviewed. Additionally, consideration should be limited to factors which examiners determine to be legitimate. FFIEC, “Interagency Fair Lending Examination Procedures,” ECF 564-9 at 22-23. For example, in July of 2007, Sandra Braunstein, Director of the Federal Reserve’s Division of Consumer and Community Affairs, testified before a congressional subcommittee about that agency’s supervisory and enforcement activities against mortgage pricing discrimination. See Braunstein Stmt., ECF 564- 6. At the outset, her testimony confirms that some of the practices the County challenges in this case—specifically, “broad discretion in pricing by loan officers or brokers” and “financial incentives for loan officers or brokers to charge borrowers higher prices—are indeed “risk factors” for mortgage discrimination.” Id. at 3. Importantly, however, she went on to explain that determining whether these risk factors materialized into discrimination requires a “closer review” of a number of product- and lender-specific factors. Id. at 3, 4. “To be accurate,” Braunstein testified, our reviews need to be based on the institution’s specific pricing policies and product offerings. Unless we take the time to understand the lender’s business and tailor our analysis accordingly, we risk either missing violations or erroneously concluding that a lender discriminated when it did not. Id. at 4 (emphasis added). Thus, to ensure the reliability of its analysis, the agency conducts “targeted pricing reviews” designed to “effectively detect discrimination.” Id. In particular, Braunstein explained, when using “statistical techniques” to perform a pricing review, we typically obtain extensive proprietary, loan-level data on all mortgage loans originated by the lender, including prime loans (i.e., not just higher-priced loans reported under HDMA). To determine how to analyze these data, we study the lender’s specific business model, pricing policies, and product offerings. With respect to product offerings, we take great care in defining the products or class of products we analyze, since each product may have different pricing that must be considered in the analysis. On the basis of our review of the lender’s policies, we determine which factors from the lender’s data should be considered. We then create a statistical model that takes into account those factors and is tailored to that specific lender. Id. (emphasis added). The County’s only response is to insist that an expert’s methodology need not “match those employed by federal regulators” and to quote a portion of the FFIEC’s Interagency Procedures stating that the Procedures are “intended to be a basic and flexible framework to be used in the majority of fair lending examinations conducted by the FFIEC agencies” and “to guide examiner judgment, not to supplant it.” Pl.’s Resp., ECF 593 at 6. What these observations overlook, however, is that it is the County’s burden to establish affirmatively that Dr. Cowan’s methodology is reliable. To be sure, the fact that no federal agency or regulator aggregates data as Dr. Cowan did when conducting fair lending examinations is not dispositive of reliability. See Smith v. Ford Motor Co., 215 F.3d 713, 720 (7th Cir. 2000) (“no single factor among the traditional Daubert list is conclusive in determining whether the methodology relied on by a proposed expert is reliable.’). But plaintiff’s failure to identify anyone—any government agency, any compliance professional in the home lending industry, any academic researcher or other commentator—who uses or endorses Dr. Cowan’s aggregated methodology,14 coupled with Braun’s testimony explaining that the Federal Reserve has found that “targeted,” 14 In response to Dr. Courchane’s report, Dr. Cowan states that Dr. Courchane herself has published “a research paper that assess (sic) pricing disparities by simultaneously aggregating across lenders and time periods.” Cowan Resp., ECF 593-3 at ¶ 40. But this wholly conclusory reference to an article the County does not even mention in its response brief falls far short of demonstrating that Dr. Courchane (or anyone else) has ever used aggregated data to perform a statistical analysis intended to detect FHA violations by specific lenders. institution- and product-specific examinations are necessary to avoid either undercounting or overcounting instances of discrimination, casts significant doubt on the reliability of his undifferentiated methodology. And indeed, Dr. Cowan’s own analysis illustrates why aggregating loan data in the way that he does yields unreliable results. For example, while four of the six opinions Dr. Cowan offers relate to putative discrimination in defendants’ loan origination practices—namely, perceived disparities in the APRs charged on, and in the overall risk profiles of, loans issued to minority borrowers and in minority neighborhoods versus those issued to white borrowers and in white neighborhoods—the data set Dr. Cowan analyzed includes a “significant” number of loans originated by lenders other than defendants, which defendants purchased post-origination. Cowan Dep., ECF 564-3 at 37:12-18. Similarly, his data set includes loans that were either originated or acquired by defendants but were foreclosed by other (non-defendant) entities. Cowan Rpt., ECF 564-1 at ¶ 18. It does not take an expert to understand that defendants could not have discriminated in the origination of loans they did not originate or in the foreclosure of loans they did not foreclose. Although Dr. Cowan was unable to quantify the number of loans in his data set that fell into one of the above categories, he acknowledged that the number was likely “significant,” raising further doubts about the reliability of his conclusions. Nothing in either the County’s or Dr. Cowan’s responsive submissions addresses this methodological flaw. It is no answer to complain that defendants’ data production was inadequate to allow Dr. Cowan to weed out loans originated or foreclosed by entities other than defendants, see Cowan Dep., ECF 564-3 at 247:11-14 (“I was provided, at best, a crippled data set that was missing a remarkable amount of data, and I did the best I could with...an impoverished data set that I had to supplement with third-party sources”), or to cast the issue as a dispute concerning the “quality of the data” that is best left for the jury, Pl.’s Resp., ECF 593 at 4. For one thing, the County apparently believes that it was appropriate for Dr. Cowan to consider loans originated by other lenders on the theory that defendants either detected or should have detected discrimination in those loans prior to purchasing them, and therefore “have knowingly undertaken the discriminatory loans.” Pl.’s Resp., ECF 593 at 8. But if there is any support for this theory of FHA liability, the County has not cited it.15 For another, Dr. Cowan’s inclusion of loan data that is not probative of defendants’ practices (as opposed to other lenders’ 15 The County offers no legal theory or any conceptual justification for Dr. Cowan’s inclusion of loans foreclosed by non-defendant entities in his analysis supporting the conclusion that defendants engaged in foreclosure discrimination. practices) in an analysis putatively designed to ascertain the impact of defendants’ practices on various populations is indeed a methodological flaw. Defendants challenge Dr. Cowan’s opinions on a number of other grounds, but the foregoing issues raise sufficient concerns about the reliability of his methodology to warrant exclusion of his testimony about: 1) defendants’ disproportionate foreclosure rates, and 2) defendants’ disproportionate issuance of higher-risk loans, and loans with higher APR rates, to minority borrowers and in neighborhoods with a high concentration of minority residents. Nevertheless, I address one additional argument defendants raise with respect to Dr. Cowan’s analysis, as it provides a segue between their Daubert arguments and their broader arguments targeting the County’s evidence of causation and damages. 3. Causation and damages Defendants argue that Dr. Cowan should not be permitted to offer any opinions on the issue of causation—specifically, that his testimony cannot be offered in support of the “robust causality” required to establish disparate impact liability— because he did not review any of defendants’ policies.16 See 16 The County raises a tepid factual challenge to this argument, pointing to Dr. Cowan’s statement at his deposition that he “looked at the comprehensive set of policies” in forming his conclusions. Cowan Dep., ECF 606-1 at 29:7-8. This is a Inclusive Communities 576 U.S. at 542 (“a disparate-impact claim that relies on a statistical disparity must fail if the plaintiff cannot point to a defendant’s policy or policies causing that disparity. A robust causality requirement ensures that ‘[r]acial imbalance ... does not, without more, establish a prima facie case of disparate impact’ and thus protects defendants from being held liable for racial disparities they did not create.”). Defendants argue that Dr. Cowan’s testimony should be excluded because it would not help the jury determine the legally relevant question, which is not simply whether the County has established a statistical disparity, but whether it has shown a statistical disparity that is caused by “a defendant’s policy or policies.” Inclusive Communities, 576 U.S. at 542. See also Daubert 509 U.S. at 591 (expert testimony must “assist the trier of fact to understand the evidence or to determine a fact in issue.”) (quoting Fed. R. Evid. 702). While defendants correctly characterize the relevant legal question, the fact that Dr. Cowan’s opinions do not wholly answer it is not a basis for exclusion. See Adams v. Ameritech Servs., Inc., 231 F.3d 414, 425 (7th Cir. 2000) (“the question perplexing statement, given that in next breath Dr. Cowan stated that he “didn’t look at” any of the policies he was asked about, and that he elsewhere testified that didn’t look at any servicing policies at all. Whatever Dr. Cowan may have meant by this testimony, however, he identified nothing that could be considered a “policy” of any defendant in the “Materials Relied On” section of his expert report. See ECF 564-1. before us is not whether the reports proffered by the plaintiffs prove the entire case... [n]o one piece of evidence has to prove every element of the plaintiffs’ case; it need only make the existence of any fact that is of consequence more or less probable.”) (internal quotation marks and citation omitted). Accordingly, I agree with the County that it may rely on other evidence to establish the causal link between the statistical disparity Dr. Cowan identifies (assuming arguendo that his opinions about those disparities are otherwise admissible) and defendants’ allegedly discriminatory practices.17 But my agreement on this point is undoubtedly cold comfort to the County, since the only additional evidence it points to for that purpose is Dr. Lacefield’s proposed testimony, which as I explained above is itself inadmissible. Moreover, even assuming that both experts’ opinions survived defendants’ Daubert 17 The County wisely does not contend that a jury could find causation based on Dr. Cowan’s testimony alone. Dr. Cowan explained his theory as follows: “My theory of causation is simple. Two people, similarly situated, the primary difference being that one is a minority and one is not a minority, are charged different interest rates for housing loans. This results in the minority paying hundreds or thousands of dollars more over time for a loan. It follows that the minority has less income for other purchases and in the event of an illness, accident, or other catastrophe is more likely to default. The borrower is more likely to default and is less likely to be able to satisfy the terms of a workout.” Cowan Resp., ECF 593-3 at ¶ 15. But this theory is based exclusively on loan origination conduct, which as I explained in my decision dismissing the SAC, cannot be deemed to have proximately caused the challenged foreclosures. motions, the County articulates no reasoned argument to explain how weaving their opinions together raises a triable issue of causation. But this is not the end of the County’s troubles on the causation front. As another court in this district has explained: The “robust causality” required in a disparate impact claim is distinct from the proximate cause analysis required of all claims under the FHA per City of Miami, 137 S. Ct. 1296. See Cty. of Cook, Illinois v. Wells Fargo & Co., 314 F. Supp. 3d 975, 990, 994 (N.D. Ill. 2018) (analyzing proximate cause and the robust causality requirement separately); City of Miami v. Bank of Am. Corp., 171 F. Supp. 3d 1314, 1320 (S.D. Fla. 2016) (making an “adequate showing” of proximate cause is insufficient to meet the separate “robust causality requirement” for a disparate impact claim). The former concerns whether a defendant’s conduct in an FHA suit is properly pleaded as the proximate cause of a plaintiff’s damages; the latter involves an examination of whether a defendant’s policies were properly pleaded as the cause of the discriminatory impact. See Alston v. City of Madison, 853 F.3d 901, 907-08 (7th Cir. 2017), cert. denied sub nom. Alston v. City of Madison, Wis., 138 S. Ct. 1571 (2018), reh'g denied, 138 S. Ct. 2714 (2018). Nat'l Fair Hous. All. v. Deutsche Bank Nat’l Tr., No. 18 CV 839, 2019 WL 5963633, at *17 (N.D. Ill. Nov. 13, 2019). My discussion above addressing the methodological flaws in the statistical analyses of the County’s experts is directed to the latter issue and explains why the County’s evidence is insufficient to allow a reasonable jury to find “robust causality” between defendants’ policies and any race-based statistical disparities in foreclosure rates. In the next section, I address the former issue and conclude that the record also does not raise a reasonable inference that defendants’ conduct proximately caused the injuries for which the County seeks damages. As noted at the outset of this decision, my decisions partially granting dismissal of the SAC and partially granting the County’s subsequent motion for clarification spelled out the “narrow category” of injuries for which the County had plausibly alleged the sort of causal link that would satisfy City of Miami’s proximate causation standard: “the out-of-pocket costs it claims to have incurred in processing the discriminatory foreclosures,” specifically: “additional funding for the Cook County Sheriff to serve foreclosure notices and for the Circuit Court of Cook County to process the deluge of foreclosures,” as well as “out-of-pocket costs in serving eviction notices, conducting judicial and administrative foreclosure proceedings, and registering and inspecting foreclosed properties.” Mem. Op. and Order of 03/30/2018, ECF 20 at 19-20 and Order of 08/17/18, ECF 228 at 1. Four years later, the County has come forward with no evidence that it provided any “additional funding” to the Sheriff’s Department, the Clerk of Court, or the Office of the Chief Judge (the three offices it claims suffered compensable losses as a result of defendants’ discrimination), nor does it identify a single “out-of-pocket” expense that it would not have incurred in the absence of defendants’ alleged discrimination. In fact, neither of the County’s damages experts identified any County costs that varied as a function of the number of foreclosures the County processed. Cowan Dep., ECF 573-96 at 80:17-81:3; Hildreth Dep., ECF 573-101 at 108:17-109:2. To the contrary, the County’s former Chief Financial Officer examined the County’s budget documents and found that these offices’ appropriations and expenditures remained essentially stable from 2004 through 2018——the period for which the County seeks damages—while residential foreclosures spiked and then declined. See Report of Thomas Glaser, ECF 612-1 at ¶¶ 21-23.18 In other words, the budget documents show that the 18 The County asks me to exclude Mr. Glaser’s testimony as unqualified and unreliable, but the motion is meritless. First, Mr. Glaser’s education—he holds a B.S. in finance and an MBA—and his experience working as an accountant, a high-level financial executive, and the County’s own Chief Financial Officer for over a dozen years—plainly qualify him to interpret the County’s budget documents and to opine on what they show about the County’s costs. Second, Mr. Glaser does not purport to “opine on the appropriate damages methodology” as a legal matter, nor does he offer opinions about “how to calculate damages for a nonprofit governmental organization such as the County.” Pl.’s Mot., ECF 590 at 4. Instead, he opines on the factual issue of whether the County’s budget materials reflect any increase in its appropriations or expenditures as a result of additional foreclosures between 2004-2018. See Glaser Rpt. at ¶¶ 8(a),(b). Mr. Glaser’s opinions in this connection are neither “subjective,” nor “speculative,” nor “unsupported.” Pl.’s Mot., ECF 590 at 6. Mr. Glaser generally describes the County’s annual budget and appropriations process and other processes through which County offices can request and receive additional funding County experienced no material increases in its costs as a result of increased foreclosures. Unable to controvert evidence that the appropriations and expenditures for each of these offices remained stable throughout the damages period, the County grudgingly admits as much, see Pl.’s Mot., ECF 590 at 9 (“appropriations and expenditures for each office as a whole may have remained somewhat stable”), then pivots to a newly-minted damages theory: that resources were “shifted” or “reallocated” within the affected offices to cope with additional burdens occasioned by the challenged foreclosures, and that the County suffered damages in the form of opportunity costs when these offices were “forced to utilize their limited resources to process the additional foreclosures resulting from Defendants’ discriminatory mortgage practices.” Pl.’s SJ Resp. ECF 622 at 39. But absent any evidence that the County failed to fund any program, to pursue any initiative, or to provide any service as a result of an internal redistribution of resources, no if needed during the fiscal year. He then identifies the materials he considered in forming his opinions and explains why these materials are likely to contain evidence, if any exists, of increases in the County’s costs. The County’s attack on Mr. Glaser’s methodology is short on reasoned analysis and long on bluster, challenging such trivialities as Mr. Glaser’s inability to recall how many pages of deposition testimony he had review and his failure to read the Illinois fee statute or some unidentified “critical Chancery Court documents” in preparing his report. These observations do not undermine the soundness of the methodology he describes. reasonable jury could find that it suffered any compensable injury that was proximately caused by defendants’ alleged FHA violations.19 Indeed, such injuries—like those the County previously claimed based on the alleged diminution of its tax digest but later conceded involved no corresponding decrease in its tax revenues—appear to be entirely “notional.” See Order of 12/19/2019, ECF 423 at 4. Because an FHA damages claim sounds in tort, Meyer v. Holley, 537 U.S. 280, 285 (2003), “general tort principles govern the award and calculation of damages,” and under those principles, “compensatory damages are designed to place the plaintiff in a position substantially equivalent to the one that he would have enjoyed had no tort been committed.” Cty. of Cook v. Wells Fargo & Co., 335 F.R.D. 166, 170 (N.D. Ill. 2020) (quoting Anderson Grp., LLC v. City of Saratoga Springs, 805 F.3d 34, 52 (2d Cir. 2015)). Because the County acknowledges that its ledger (including the appropriations and expenditures of the three offices it claims had additional costs) was not affected by the alleged discrimination, compensatory damages are unavailable. 19 But see Valencia v. City of Springfield, Illinois, No. 16- 3331, 2020 WL 1265421, at *6 (C.D. Ill. Mar. 16, 2020) (denying summary judgment without discussing proximate causation of the organizational plaintiff’s FHA claim seeking a portion of its employees’ salaries on the theory that their time was diverted from the organization’s mission to combat the defendants’ alleged discrimination). Further, whatever the merits of the County’s “average cost methodology” in principle, one thing is clear: the County’s cost tabulations ignore the damages limitations I concluded were necessary to ensure that the County could recover only for injuries that satisfied City of Miami’s proximate causation analysis. As the County frankly admits, its cost averaging methodology “captured all the County’s costs, including overhead and shifted resources,” that were associated in any manner with the challenged foreclosures. Pl.’s Resp., ECF 622 at 43 (emphasis added). In fact, fully a quarter of the damages it seeks are overhead costs that include administrative and support services, and some portion of these—though the County cannot say how much—is entirely unrelated to foreclosure processing: commissioner salaries, premiums for employee health insurance and pensions, information technology, facilities management and depreciation, supplies, and the like. Conspicuously, the County’s own expert refers to these as “indirect costs” related to foreclosure processing. Def.’s L.R. 56.1 Stmt., ECF 577 at ¶ 98. Yet, City of Miami “requires some direct relation between the injury asserted and the injurious conduct alleged.” 137 S. Ct. at 1306 (internal quotation marks and citation omitted). To award the County damages for such overhead costs would surely violate this principle. The County champions its cost averaging methodology as an alternative to itemizing the incremental costs it incurred to process the challenged foreclosures—an exercise it admits would be “administratively unfeasible, and virtually impossible[.]” Pl.’s Resp., ECF 622 at 43. But this only underscores that the damages the County seeks cannot be reconciled with the idea that proximate causation turns, in part, on “an assessment of what is administratively possible and convenient.” City of Miami, 137 S. Ct. at 1306 (citation omitted). See also Kemper v. Deutsche Bank AG, 911 F.3d 383, 392 (7th Cir. 2018) (“we use ‘proximate cause’ to label generically the judicial tools used to perform an inquiry that ultimately reflects ideas of what justice demands, or of what is administratively possible and convenient”) (some internal quotation marks and citations omitted). Lacking any feasible way to identify and segregate overhead costs that are directly related to foreclosure processing, the County employed a damages model that would make defendants responsible for a portion of costs entirely unrelated to their conduct. The County may not simply overlook City of Miami’s directness requirement in exchange for administrative convenience. It is true that the Supreme Court declined, in City of Miami, “to draw the precise boundaries of proximate cause under the FHA.” 137 S. Ct. 1306. But even its broad brush strokes in that case make clear that a significant portion of the damages the County claims amount to distant “ripples of harm” far removed from defendants’ conduct. Id. at 1299. Accordingly, they are not compensable under the statute. IIl. For the foregoing reasons, defendants’ motions to exclude the expert testimony of Gary Lacefield and the liability opinions of Charles Cowan are granted, as is their motion for summary judgment. The County’s motion to exclude the testimony of Thomas Glaser is denied. ENTER ORDER: Elaine E. Bucklo United States District Judge Dated: February 10, 2022 58