In Re: RFC and RESCAP Liquidating Trust Litigation, No. 0:2013cv03451 - Document 2678 (D. Minn. 2017)

Court Description: ORDER denying Defendants' requests [Doc. Nos. 2311 & 2376] with respect to Plaintiffs' AVM methodology and appraisal-based breaches. (Written Opinion). Signed by Judge Susan Richard Nelson on 06/30/17. (MJC)
Download PDF
In Re: RFC and RESCAP Liquidating Trust Litigation Doc. 2678 UNITED STATES DISTRICT COURT DISTRICT OF MINNESOTA In Re: RFC and RESCAP Liquidating Trust Actions Civil File No. 13-3451 (SRN/HB) ORDER Order Regarding Plaintiffs’ Amended AVM Methodology ________________________________________________________________________ SUSAN RICHARD NELSON This order addresses a dispute involving the ability of Plaintiffs to utilize a type of Automated Valuation Model (“AVM”) different than previously identified, and to add new AVM-based and appraisal-based breach allegations. The parties’ positions are set forth in the letters of March 14, 2017 [Doc. Nos. 2311 & 2317], Plaintiffs’ Memorandum in Support of Reliance on Amended AVM Breach Lists [Doc. No. 2361], the Declaration of Sascha N. Rand Regarding Plaintiffs’ AVM Breach List Disclosures [Doc. No. 2362], and the Opt-In Defendants’ Response Memorandum [Doc. No. 2376]. I. BACKGROUND A. AVM Methodology In support of Plaintiffs’ position with respect to their use of an amended AVM methodology, Plaintiffs have submitted the Declaration of Sascha N. Rand, a partner at Quinn Emanuel Urquhart & Sullivan, LLP, counsel for Plaintiffs. Ms. Rand has experience in representing clients with claims involving the sale and securitization of 1 residential mortgage loans. (Rand Decl. ¶ 4 [Doc. No. 2362].) In particular, she has represented the Federal Housing Finance Authority (“FHFA”) in actions against various financial institutions and served as trial counsel in FHFA v. Nomura Holding America, Inc. See Nomura, 104 F. Supp. 3d 441 (S.D.N.Y. 2015). Ms. Rand was responsible for prosecuting FHFA’s appraisal breaches in Nomura, which involved work on the AVM analysis. (Rand Decl. ¶ 4.) Ms. Rand explains that a loan-to-value ratio compares the market value of a residential property that provides collateral for the loan against the actual loan amount. (Id. ¶ 5) (citing Nomura, 104 F. Supp. 3d at 497). Once the loan-to-value ratio nears 100% or more, the home is worth as much as, or less than, the amount of the mortgage, which can harm a borrower’s ability and incentive to continue making mortgage payments. (Id.) An inflated appraisal may serve to increase the “value” element of the loan-to-value ratio, and, as a result, depress the loan-to-value ratio. (Id. ¶ 6.) This may give a loan the appearance of complying with a contractual representation regarding the loan-to-value ratio when, in fact, the loan is not in compliance. (Id.) (citing Nomura, 104 F. Supp. 3d at 517-18.) Ms. Rand identifies two methods of determining the value of a loan-to-value ratio. (Id. ¶¶ 7-8.) First, the value may be determined through an appraisal conducted by a licensed residential appraiser. (Id. ¶ 7.) The court in Nomura found evidence from which to infer that residential appraisers were under substantial pressure to inflate the value estimate of residential mortgage loans in the time period in question. (Id.) (citing Nomura, 2 104 F. Supp. 3d at 517-18.) Second, the value of a loan-to-value ratio may be determined by the use of AVMs–“data-driven analytical models used to predict the market value of real estate properties.” (Id. ¶ 8.) Although the supporting methodology of an AVM may vary, an AVM generally “compares the characteristics of an at-issue property (the ‘subject’ property) to the characteristics of comparable properties in the geographical area of the subject property that were sold at or around the relevant point in time (the ‘comparable sales’).” (Id.) Experts then analyze the property characteristics and comparable sales, often using a regression analysis, to predict the market value of the property at the relevant time. (Id.) As to AVM methodologies, litigants may use commercially-available, off-the-shelf “black-box” AVMs, offered by real estate data and analysis firms and used in the residential mortgage loan underwriting industry (Id. ¶ 9.) For a nominal fee, commercial AVMs can quickly return property value estimates based only on a property address. (Id. ¶ 10.) But, Ms. Rand states that because commercial AVMs are proprietary, the AVM providers generally do not publicly reveal information about the AVMs’ design, operation, limitations, analyses, algorithms, coding, data, or validation. (Id.) Ms. Rand states that “commercial AVM providers are generally unwilling to produce individuals knowledgeable about the inner workings of a commercial AVM to answer technical questions, or to give deposition testimony.” (Id.) (citing Nomura, 104 F. Supp. 3d at 503.) Alternatively, litigants can hire an expert with his or her own AVM model or access to a third party’s AVM model that can be adapted for the use of a particular case. (Id. ¶ 11.) 3 Ms. Rand asserts that while such models may not have the same black-box limitations of commercial AVMs, “they typically require substantial time and effort to customize and implement for use in litigation and, to the extent they have not been used commercially and are not industry-recognized, can be more difficult to validate.” (Id.) Ms. Rand identifies the Greenfield AVM (“GAVM”) as one such non-commercial AVM, developed by Dr. John Kilpatrick in connection with work performed for his consulting firm, Greenfield Advisors, LLC. (Id. ¶ 12.) As noted, in contrast to commercial AVMs, securing the data necessary to implement the GAVM in a given case takes considerable time. (Id.) Instead of simply using an address to generate a property value estimate, the GAVM requires the collection of data–sometimes collected manually–about the subject property from, among other sources, the appraisal and other mortgage-related documents, and is a “process [that] can take many months.” (Id. ¶ 13.) In the Nomura litigation, Dr. Kilpatrick used the GAVM as part of his methodology to identify inflated appraisals. (¶ 15.) He used the GAVM to provide an “objective retrospective assessment of the value of the relevant properties,” and an appraisal review to “support[ ] the reliability of the GAVM and evaluate[ ] whether the appraisal was properly perofrmed and whether the appraisal opinion was supported.” (Id. ¶ 16) (citing Nomura, 104 F. Supp. 3d at 503.) B. AVM Disclosures in this Litigation Here, prior to the consolidation of the cases, and as part of Plaintiffs’ Rule 26(a) disclosures, Plaintiffs were required to provide Defendants with two types of lists 4 identifying the loans alleged to be defective, within a 28-day period: (1) AVM breach lists, identifying loans for which an AVM, retrospectively run by Plaintiffs, produced property values different from those used at origination; and (b) re-underwriting defect lists, listing allegedly breached loans that were identified by Plaintiffs’ re-underwriting vendors. (Rand Decl. ¶ 17; Defs.’ 3/14/17 Letter at 2.) Particularly in light of the short time table, Plaintiffs used a commercial, black-box AVM methodology to prepare the initial AVM breach lists. (Rand Decl. ¶ 17.) The original consolidated case management order (the “CMO”) acknowledged that Plaintiffs had produced initial re-underwriting disclosures, including an AVM breach list, a re-underwriting defect list, and a loan list. (CMO, § II(A) [Doc. No. 277].) The CMO also recognized that Defendants sought supplementation of these disclosures and directed Plaintiffs to supplement their disclosures on a rolling basis and provide a status report at the monthly case management conferences. (Id.) In addition, the CMO permitted Plaintiffs to modify and supplement their disclosures thereafter “(1) on the basis of newly acquired information; (2) or for good cause shown, with leave of Court, on the basis of information within their possession or control.” (CMO, § II(B).) The CMO required Defendants to serve on Plaintiffs rebuttals to the re-underwriting disclosures, but stated that the rebuttals “need not respond to the AVM Breach List.” (Id., § III(A)-(B).) Plaintiffs served several sets of AVM results during the course of this litigation. (See Pls.’ Letter of 3/14/17 at 3 [Doc. No. 2317].) Plaintiffs contend that their lists and supplements stated that their expert analysis was ongoing and the AVM results were 5 therefore subject to change. (Id.) As contemplated by the CMO, Defendants have not served rebuttal reports to the AVM lists. (Id.) In the summer of 2016, however, Plaintiffs assert that based on recent developments in an RMBS-related case, U.S. Bank, National Association v. UBS Real Estate Securities Inc., No. 12-7322 (S.D.N.Y.) (“MARM”)1, they considered using a new expert AVM methodology in this consolidated action, (Pls.’ Letter of 3/14/17 at 2), as they stated at the June 2016 case management conference. (H’rg Tr. of 6/23/16 at 127-28 [Doc. No. 127-28].) Because the MARM defendants had criticized the black-box AVM methodology in that case, arguing that it was unavailable for their examination, Plaintiffs here anticipated that Defendants would similarly challenge their AVM methodology. (Id.) Therefore, they raised the possibility of utilizing “a more transparent AVM . . . where the parties could open the box” and test the underlying algorithm. (Id.) The prospect of Plaintiffs implementing a new AVM methodology arose several times thereafter. In June 2016, after Dr. Kilpatrick and Greenfield Advisors confirmed that they could implement the GAVM for Plaintiffs in this consolidated action, (Rand Decl. ¶ 29), Plaintiffs informed Defendants, prior to the July 2016 case management conference, that they had decided to implement the GAVM methodology. (See H’rg Tr. of 8/18/16 at 48 [Doc. No. 1744].) The subject also arose during the August 2016 case 1 U.S. Bank brought the suit in its capacity as the trustee of several Master Adjustable Rate Mortgages Trusts. See MARM, 205 F. Supp. 3d 386 (S.D.N.Y. 2016). The plaintiffs in MARM were represented by one of the same law firms representing Plaintiffs here, Quinn Emanuel. 6 management conference, (id.), a November 2016 meet and confer, (Pls.’ 3/14/17 Letter at 3), and at the November 2016 case management conference. (H’rg Tr. of 11/17/16 at 6364 [Doc. No. 1970].) At the November 2016 case management conference, Plaintiffs reported to the Court that they were preparing new lists based on the new methodology and they planned to serve most or all of the lists by the December 2016 case management conference. (Id.) Plaintiffs state that as to the 15 cases in this consolidated proceeding, they served new AVM lists in the 11 remaining sampling cases on December 13, 2016, and in three of the four non-sampling cases in February 2017. (Pls.’ 3/14/17 Letter at 1 n.1.) Plaintiffs maintain that the new AVM lists add approximately 14 new AVM breaches per case, but subtract larger numbers of breaches. (Id. at 4.) C. Parties’ Positions Citing the CMO, Defendants argue that Plaintiffs should be precluded from advancing their AVM breach claims, appraisal-based breach claims, and otherwise revising their breach allegations, absent a showing of newly acquired information or good cause. (Defs.’ 3/14/17 Letter at 1-2; Defs.’ Mem. at 1-3 [Doc. No. 2376].) They contend that no information was newly acquired, nor have Plaintiffs demonstrated good cause. (Id.) Moreover, they contend that even if the number of loans is reduced by Plaintiffs’ new disclosures, Defendants will still have to perform the work of contract-matching on those new loans “from scratch.” (Defs.’ Mem. at 8-9.) Plaintiffs, however, argue that the CMO is inapplicable to the AVM disclosures. 7 But even if the CMO applies, they contend that they have demonstrated good cause. II. DISCUSSION As Plaintiffs note, in the past, other courts have permitted experts to use commercial, black-box, off-the-shelf AVMs in determining whether appraisals were inflated. See Assured Guar. Mun. Corp. v. Flagstar Bank, FSB, 920 F. Supp. 2d 475, 507 (S.D.N.Y. 2013); Spears v. First Am. eAppraiseIT, No. 5-08-00868-RMW, 2014 WL 4647679, at *10 (N.D. Cal. Sept. 16, 2014). More recently, however, courts have permitted the use of a non-commercial AVM methodology. See Nomura, 104 F. Supp. 3d at 599; Mass. Mut. Life Ins. Co. v. DB Structured Prods, Inc., No. 11-30030-MGM, 2015 WL 2130060, at * 9-10 (D. Mass 2015). Plaintiffs here continued to use a commercial AVM approach until the spring and summer of 2016 when two “interrelated developments” caused Plaintiffs to reconsider their approach. (Rand Decl. ¶ 24.) First, in the March 2016 Daubert arguments in MARM, the defendants attacked the plaintiffs’ use of a black-box AVM. This concerned Plaintiffs here, despite the admissibility of such a methodology in Assured, that their reliance on black-box AVMs would be subject to similar challenges. (Id. ¶ 25.) Second, after Plaintiffs worked with a commercial AVM vendor and possible expert witnesses, they determined that “contrary to prior expectations, they would be unable to proffer an expert witness who would be able to understand and testify satisfactorily about the methodology, underlying data, and reliability of the commercial AVM Plaintiffs had utilized thus far.” (Id. ¶ 26.) While Plaintiffs dispute whether the original CMO applies to the type of 8 methodology changes at issue here, (Pls.’ Letter of 3/14/17 at 5), assuming that it does, the Court finds that Plaintiffs have satisfied the CMO’s good cause requirement. Plaintiffs have demonstrated that the accepted use of a commercial AVM to determine loan-to-value ratio changed between 2014 and 2015, when Plaintiffs served their initial AVM breach lists, and mid-2016, when the MARM defendants attacked the use of a commercial AVM. By mid-2016, Plaintiffs believed that they would not be able to offer an expert to support the reliability of their commercial AVM, particularly as the more customized GAVM had survived legal challenges in other litigation. (Rand Decl. ¶ 25-28.) Defendants argue that nothing prevented Plaintiffs from using the “new” methodology at the outset of this litigation, and they question why Plaintiffs switched course in the summer of 2016, when the MARM decision was not issued until September 2016. (Defs.’ Mem. at 2 [Doc. No. 2376].) But, as Plaintiffs have explained, it appeared that the black-box methodology–notably, a faster and simpler approach–was generally accepted by courts. Plaintiffs became aware of the limitations of that methodology upon learning of the MARM defendants’ Daubert arguments in March 2016, even though the MARM decision was issued six months later. As to why it took Plaintiffs “half a year to actually switch here,” (see id. at 4), Ms. Rand attests to her understanding that “given the number of cases at issue here, Dr. Kilpatrick was not in [a] position to take on this matter until the Summer of 2016.” (Rand Decl. ¶ 28.) Also, when considering the time line here, Defendants were not required to respond to Plaintiffs’ earlier AVM breach lists, had not put them at issue in fact discovery, and, at the time that Plaintiffs broached the subject 9 of utilizing an updated AVM methodology, expert discovery was not scheduled to begin for several months. The Court finds that because of recent legal developments concerning the acceptance of GAVM methodology and the rejection of black-box, commercial AVM methodology, Plaintiffs have demonstrated good cause for amending their AVM methodology. See, e.g., Pumpco, Inc. v. Schenker Int’l, Inc., 204 F.R.D. 667, 668 (D. Colo. 2001) (finding good cause to extend a scheduling order deadline based on information learned through discovery and a recent change in the law); Jo Ann Howard & Assocs, P.C. v. Cassity, No. 9-CV-01252, 2014 WL 6607077, at *4 (E.D. Mo. Nov. 19, 2014) (noting that good cause requires a change in “circumstance, law, or newly discovered facts.”) While Defendants claim they will be prejudiced by Plaintiffs’ use of the GAVM methodology, the Court disagrees. Again, under the original CMO, Defendants were not required to respond to Plaintiffs’ earlier AVM Breach Lists, (see CMO § III(B)), and Plaintiffs assert that Defendants have not responded to the calculations in the lists. (Pls.’ Mem. at 6 [Doc. No. 2361].) Moreover, while the Court acknowledges that Defendants have spent time and money addressing 72 AVM-based breach allegations on 22 loans, (Defs.’ Letter of 3/14/16 at 7-8), it appears that the number of new breaches identified through the new methodology is relatively small. (See Pls.’ Letter of 3/14/17 at 4 n.3 & 8.) In addition, the updated AVM methodology and revised AVM breach lists were served six months before the start of expert discovery in the sampling cases, and eight months 10 before Defendants are required to serve their opposition to Plaintiffs’ AVM expert reports. For all of the foregoing reasons, the Court finds that good cause, and a lack of significant prejudice to Defendants, exists here with respect to the use of the GAVM methodology. As to Defendants’ related argument concerning new appraisal-based breach allegations, Defendants seek to limit Plaintiffs to “using the appraisal reviews to support AVM-based breaches that were already disclosed in the CMO-mandated AVM Breach Lists.” (Defs.’ Mem. at 11.) Given the Court’s finding of good cause with respect to Plaintiffs’ GAVM methodology, however, Defendants’ request is denied. THEREFORE, IT IS HEREBY ORDERED THAT: 1. Defendants’ requests [Doc. Nos. 2311 & 2376] with respect to Plaintiffs’ AVM methodology and appraisal-based breaches are DENIED. Dated: June 30, 2017 s/Susan Richard Nelson SUSAN RICHARD NELSON United States District Judge 11