In Re: RFC and RESCAP Liquidating Trust Litigation
Filing
374
ORDER granting in part and denying in part without prejudice #154 Motion to Approve Use of Statistical Sampling (Written Opinion). Signed by Judge Susan Richard Nelson on 04/16/15. (SMD)
UNITED STATES DISTRICT COURT
DISTRICT OF MINNESOTA
In Re: RFC and ResCap Liquidating
Trust Litigation
_________________________________
This document relates to:
Residential Company, LLC v. Guaranty
Bank, No. 13-cv-3450 (MJD/FLN)
Residential Funding Company, LLC and
ResCap Liquidating Trust v. Community
West Bank, N.A., No. 13-cv-3468 (JRT/JJK)
Residential Funding Company, LLC and
ResCap Liquidating Trust v. Colonial
Savings, F.A., No. 13-cv-3474 (JNE/TNL)
Residential Funding Company, LLC v. First
Guaranty Mortgage Corporation, No. 13-cv3475 (RHK/HB)
Residential Funding Company, LLC and
ResCap Liquidating Trust v. Provident
Funding Associates, L.P., No. 13-cv-3485
(SRN/TNL)
Residential Funding Company, LLC and
ResCap Liquidating Trust v. PNC Bank,
N.A., as successor in interest to National
City Mortgage Co., NCMC Newco, Inc., and
North Central Financial Corporation, No. 13cv-3498 (JRT/BRT)
Residential Funding Company, LLC v.
Impac Funding Corp., No. 13-cv-3506
(JNE/SER)
Residential Funding Company, LLC v.
Hometown Mortgage Services, Inc., No. 13cv-3509 (PAM/HB)
Residential Funding Company, LLC and
ResCap Liquidating Trust v. Universal
Civ. No. 13-3451 (SRN/JJK/HB)
ORDER
American Mortgage Company, LLC, No. 13cv-3519 (SRN/JSM)
Residential Funding Company, LLC v. Wells
Fargo Bank, N.A. f/k/a Wachovia Mortgage
Corporation, First Union National Bank, and
First Union Mortgage Corporation, No. 13-cv3521 (ADM/LIB)
Residential Funding Company, LLC v. BMO
Harris Bank, N.A. d/b/a M&I Bank, FSB, No.
13-cv-3523 (JNE/FLN)
Residential Funding Company, LLC v.
SouthTrust Mortgage Corp. and Wells Fargo
Bank, N.A., No. 13-cv-3524 (DSD/HB)
Residential Funding Company, LLC v. Wells
Fargo Financial Retail Credit, Inc., No. 13-cv3525 (SRN/JSM)
Residential Funding Company, LLC and
ResCap Liquidating Trust v. Standard Pacific
Mortgage, Inc., No. 13-cv-3526 (JRT/JJK)
Residential Funding Company, LLC and
ResCap Liquidating Trust v. National Bank of
Kansas City, No. 13-cv-3528 (ADM/TNL)
Residential Funding Company, LLC v. DB
Structured Products, Inc. and MortgageIT, Inc.,
No. 14-cv-143 (ADM/TNL)
Residential Funding Company, LLC v. CTX
Mortgage Company, LLC, No. 14-cv-1710
(DSD/TNL)
Residential Funding Company, LLC v. Home
Loan Center, Inc., No. 14-cv-1716 (DWF/JJK)
Residential Funding Company, LLC v.
Decision One Mortgage Company, LLC and
HSBC Finance Corporation, No. 14-cv-1737
(MJD/JSM)
Residential Funding Company, LLC v.
2
E-Loan, Inc., No. 14-cv-1739 (PAM/JJK)
Residential Funding Company, LLC v. Rescue
Mortgage, Inc., No. 14-cv-1740 (PJS/TNL)
Residential Funding Company, LLC v.
American Mortgage Network, LLC f/k/a
American Mortgage Network, Inc., d/b/a
Vertice and AMNET Mortgage LLC, f/k/a
AMNET Mortgage, Inc., f/k/a American
Residential Investment Trust, Inc. and Wells
Fargo Bank, N.A., No. 14-cv-1760 (PJS/TNL)
ResCap Liquidating Trust v. Freedom
Mortgage Corporation, No. 14-cv-5101
(MJD/HB)
_________________________________
SUSAN RICHARD NELSON, United States District Judge
In their Motion to Approve Use of Statistical Sampling (Doc. No. 154, Sampling Mot.),
Plaintiffs Residential Funding Company, LLC, and the ResCap Liquidating Trust (collectively
“RFC”) seek an order in limine approving their proposed sampling methodology to determine a
breach rate for groups of loans sold to RFC by the Defendants in 23 cases where the total number
of loans exceeded 500 in an individual case. For the reasons that follow, the motion (Doc.
No. 154), is GRANTED IN PART, and DENIED IN PART WITHOUT PREJUDICE as
further stated below.
RFC’s expert, Dr. Karl Snow, has designed a sampling protocol that is described in his
expert declaration submitted in support of this motion. (Doc. No. 157-11, Expert Decl. of Karl
Snow, Phd (“Snow Decl.”) ¶¶ 37-41.) According to his sampling protocol, Dr. Snow will draw a
random sample of 150 loans and a backup sample of 100 loans from a population of loans sold
by a defendant that experienced losses, or expected losses, greater than $500. He will then
conduct statistical tests to confirm the representatives of the sample. After the loans in the
3
sample are re-underwritten to determine whether any representations and warranties made by a
defendant were breached, Dr. Snow will extrapolate the sample-based breach rates to the
population from which the sample was drawn. Dr. Snow opines that a sample of 150 loans is
sufficiently large to identify a breach rate for each population of loans at issue in these 23 cases
with a 95% confidence interval and a margin of error of 8 percentage points. (Id. at ¶ 39.)
Based on Dr. Snow’s declaration, in its memorandum in support of the instant motion,
RFC requests the following relief:
The Court should (i) hold that Plaintiffs may attempt to establish liability and
determine damages by use of a statistically valid random sample of loans in
substantially the manner set forth in the declaration of Dr. Karl N. Snow
submitted with plaintiffs’ motion; and (ii) admit into evidence the expert
testimony of Dr. Karl N. Snow regarding statistical sampling pursuant to Federal
Rule of Evidence 702.
(Doc. No. 156, Pls.’ Mem. of Law in Supp. of Mot. to Approve Use of Statistical Sampling
(“Pls.’ Mem.”) 32; Doc. No. 154, Sampling Mot.)
The Defendants object to the entry of such an order. First, Defendants argue that RFC
has not demonstrated that Dr. Snow’s proposed sampling methodology would reliably generate
relevant evidence because RFC failed to offer any explanation of how it intended to use the
proposed breach rates. Accordingly, Defendants argue that the Court should not rule on whether
RFC may use Dr. Snow’s method of determining breach rate through statistical sampling to
establish liability or damages until the Court knows the use for which this evidence is to be put
and until Defendants have the opportunity to object to its admission. Second, the Defendants
argue that the proposed sampling is premature because, at least by Defendants’ theory of the
case, RFC ultimately will have to prove breaches on a loan-by-loan basis, and Dr. Snow’s
sampling exercise does not purport to provide loan-by-loan proof of breach. At the very least, if
the Court endorses the use of statistical sampling, Defendants assert that the Court should not
4
prejudge the language of the contracts as to the issue of proof of breach. Third, Defendants
argue that they will demonstrate “at the appropriate time” that, to recover for any losses
occasioned by a Defendant’s loans that RFC sold into a particular securitization, RFC must
prove that there were specific loans backing the securitization that breached the contractual
representations and warranties and that those breaches caused settlement loss. According to
Defendants, Dr. Snow’s sampling protocol only draws its 150-loan samples from the entire
population of loans with losses sold by any given Defendant, regardless of the securitization in
which such loans were placed, and thus, provides no information about the number of
purportedly breaching loans within any particular securitization. As a result, Defendants argue
that the sampling protocol will be a waste of time and resources and will not yield any
admissible evidence. Fourth, the Defendants contend that numerous other aspects of Dr. Snow’s
protocol cannot be evaluated at this time including: (1) whether Dr. Snow will reliably pick his
loan population and samples; (2) whether his use of loans from his back-up samples will bias the
sample sent for re-underwriting; (3) whether the re-underwriting will focus on relevant criteria;
(4) whether the extrapolation of results will be reliably performed; and (5) whether Dr. Snow’s
samples will contain time-barred loans sold prior to May 14, 2006. (See Doc. No. 255, Defs.’
Joint Opp’n to Pls.’ Mot. to Approve Use of Statistical Sampling (“Defs.’ Opp’n”), passim.)
Notably, Defendants do not contest the fact that the statistical analysis proposed by Dr. Snow can
be used to estimate breach rates in populations of mortgage loans, nor do they raise any objection
to Dr. Snow’s expert qualifications.
In RFC’s Reply Memorandum, RFC suggests the following alternative to its initial
request for an order approving statistical sampling as a method to prove liability and damages
and admitting Dr. Snow’s expert testimony “into evidence” at the outset of the case:
5
Alternatively, the Court should hold the protocol described in Dr. Snow’s
declaration is scientifically valid and admissible in this case for the purpose of
identifying a random, representative sample of loans from the indicated
population, and that if the protocol is implemented as Dr. Snow indicates, the
Sampling Defendants may not challenge on Daubert grounds that a 150-loan
sample is sufficiently large to identify a breach rate with a margin of error at the
95% confidence level of no more than +/- 8 percentage points for binary
questions.
(Doc. No. 333, Pls.’ Reply Mem. in Further Supp. of Mot. to Approve Use of Statistical
Sampling (“Pls.’ Reply”) at 30.) This alternative proposal simply recognizes the mathematical
principles underlying Dr. Snow’s proposed statistical sampling methodology. It does not include
any determination about whether – or how – RFC may use the results of the statistical sampling
exercise at trial. Nor does it require this Court to rule on the admission of the expert testimony
of Dr. Snow at trial. RFC derived this alternative approach from the statistical sampling order
issued by Bankruptcy Judge Martin Glenn in the four parallel cases pending in the United States
Bankruptcy Court Southern District of New York. In re Residential Capital, LLC, No. 14-ap7900-mg, Doc. No. 56 at 5 (Bankr. S.D.N.Y. Jan. 13, 2015) (“Resolving the sampling
methodology and sample selection now will substantially aid in the prompt and efficient
administration of these cases without prejudicing the parties’ rights.”)
At the hearing on March 31, 2015, Defendants’ counsel stated that Defendants did not
object to the substance of RFC’s alternative proposal so long as it was made clear that
Defendants maintained their right to challenge various ways in which the sampling protocol is
implemented and applied. 1 Defendants also wish to preserve each of the legal challenges,
1
Defense counsel stated that: “If the question is whether drawing 150 loan sample from a
population at random will . . . allow you to say that whatever result applies to the – that
population as a whole with a 95 percent confidence level at a maximum plus or minus 8 percent
margin of error, then the answer is, yes, we don’t disagree with that.” (Doc. No. 359, Tr. of
March 31, 2015 Hr’g 143:2-8.)
6
described above, that they raised in response to RFC’s original request for a determination that
RFC may use evidence of a breach rate to prove liability and damages. Defendants did,
however, suggest that it would be more efficient for the Court to wait until much later in the case
to make any ruling on statistical sampling.
The Court will issue an order granting the alternative relief sought by RFC. There are
significant benefits that can be derived from this early ruling on the use of the sampling protocol
to identify the loan sample size and the method of identifying a breach rate for a defined
population of loans sold by a defendant to RFC. The re-underwriting process that is involved in
sampling 150 loans is expensive and time-consuming. If it turned out at the end of fact and
expert discovery that the sampling methodology was not valid (i.e., a sample of 150 loans was
not sufficiently large to identify a breach rate or the margin of error is too large) then Plaintiffs
would be forced to resample, simply because RFC’s expert had not come up with a statistically
valid protocol. Early disclosure by RFC of the identity of the sample loan files selected by Dr.
Snow and the results of his tests for representativeness will assist the Defendants in deciding
whether to use different sampling methodologies and different samples to conduct their own
tests. And prompt disclosure of the results of the actual re-underwriting of the sample loans will
not only provide the parties with important information about the merits of the case but will also
allow adequate time for defendants to conduct discovery about the re-underwriting process in
preparation for trial.
Whether or not any early Court order about statistical sampling is issued, it is inevitable
that, given the size of the loan pools in issue, statistical sampling of some kind will be done.
And defendants will be pressing, as they already have, to find out the identity of each loan file
sampled, the nature of any claimed defect in a loan for which breach is claimed, the process by
7
which loan files were actually chosen for sampling, the process and details of the sample reunderwriting and the results obtained. Defendants will need to obtain this information not only
to prepare their defense at trial, but also to assist them in deciding whether to conduct their own
sampling or their own re-underwriting of the RFC’s samples. RFC, in turn, will seek to discover
a similar array of information from the Defendants. The establishment of a statistical sampling
disclosure schedule will assist the parties and the Court in managing such discovery in this very
complex litigation. The Court will manage the statistical sampling process by requiring the
parties to exchange information on a statistical sampling disclosure schedule described below.
This Court recognizes that three judges in this District decided that RFC’s motions for an
order on the admissibility of the sampling protocol were premature and denied the motions
without prejudice in individual cases before these matters were consolidated for pretrial
administration. Residential Funding Co. v. PNC Bank, N.A., No. 13-cv-3498 (JRT/JSM) (D.
Minn.) (Doc. No. 117, Dec. 17, 2014); Residential Funding Co. v. First Guaranty Mortgage
Corp., No. 13-cv-3475 (RHK/HB) (D. Minn.) (Doc. No. 69, Dec. 12, 2014); Residential Funding
Co. v. Golden Empire Mortg., Inc., No. 13-cv-3466 (PJS/JJK) (D. Minn.) (Doc. No. 88, Nov. 26,
2014). However, in those pre-consolidation sampling motions, RFC sought a determination that
it could use expert testimony and analysis of statistical sampling to prove liability and damages
at trial, which is what RFC originally sought when it filed the sampling motion now before this
Court. 2 This Court will not make any determination at this stage of the case about whether – or
2
In their pre-consolidation rulings, Judges Tunheim, Kyle, and Schiltz all recognized that
determining the admissibility of expert evidence prior to the completion of discovery was
premature and would not be appropriate. See Residential Funding Co. v. PNC Bank, N.A., No.
13-cv-3498 (JRT/JSM), Doc. No. 117 at 2 (“Although the Court may ultimately agree with
Plaintiffs that they may prove their case at trial using statistical sampling, at this point, the Court
concludes that it is premature to make a determination about the appropriateness of statistical
(Footnote Continued on Following Page)
8
how – RFC may use statistical sampling at trial. Also, to the extent those judges were concerned
that RFC’s previous motions involved possible limitations on the Defendants’ right to conduct
discovery outside of the sample loans, RFC has now disavowed any such attempt to limit
discovery. Compare Residential Funding Co. v. Golden Empire Mortg., Inc., No. 13-cv-3466
(PJS/JJK), Doc. No. 88 at 3 (“Plaintiffs apparently hope to obtain an early ruling on this issue in
order to limit the scope of discovery”), with Pls.’ Mem. 10 (“Plaintiffs do not seek by this motion
to restrict Defendants’ ability to obtain discovery to which they are entitled, and this motion, if
granted, would not do so.”). Here, as noted above, RFC has significantly narrowed the request
for relief in these 23 cases and seeks only a limited determination that Dr. Snow will not be
challenged on the grounds that he is not qualified as an expert on statistical sampling (which
Defendants have not challenged), and a determination that Dr. Snow’s proposed sampling
protocol would yield a statistical result at a particular confidence level with a particular margin
of error for the populations sampled (a proposition Defendants indicated they would not dispute,
see supra note 1).
As they did in the companion cases before Judge Glenn in the Bankruptcy Court for the
Southern District of New York, Defendants in these cases have preserved their rights to object in
the future to any use by RFC of the sample and its analysis to prove liability or damages. (See In
re Residential Capital, LLC, No. 14-ap-7900-mg, Doc. No. 66 ¶¶ 14-15 (Bankr. S.D.N.Y. Feb.
10, 2015).) This includes, for example, the Defendants’ argument that the breach rate derived
(Footnote Continued from Previous Page)
sampling.”); Residential Funding Co. v. First Guaranty Mortgage Corp., No. 13-cv-3475
(RHK/HB), Doc. No. 69 at 1-2 (agreeing with other courts that an early decision on the
admissibility of expert evidence was premature); Residential Funding Co. v. Golden Empire
Mortg., Inc., No. 13-cv-3466 (PJS/JJK), Doc. No. 88 at 3 (“[T]he Court will not hear Daubert or
summary-judgment motions until discovery is completed, unless a magistrate judge recommends
otherwise and the Court agrees with that recommendation.”).
9
from the sampling has no relevance because the contract in issue requires RFC to prove breach
on a loan-by-loan basis. Nor does this order foreclose Defendants’ argument that the overall
breach rate derived by Dr. Snow’s proposed analysis is not relevant because in order to recover,
RFC must prove that there were specific loans backing a particular securitization that breached
the contractual representations and warranties, and that those breaches caused loss through the
settlement of a claim arising out of that securitization. If relevant to RFC’s theories of liability
and damages, the breach rate may be admissible (assuming the methodology was properly
applied) and may be accorded such weight as appropriate. See In re Residential Capital, LLC,
No. 14-ap-7900-mg, Doc. No. 66 ¶ 15 (Bankr. S.D.N.Y. Feb. 10, 2015). Further, Defendants
have preserved their rights to object to the implementation of the sample protocol. This includes
the rights of the Defendants to challenge the following:
(1) The applicability of [RFC’s] Sampling Protocol to loans outside the
Population;
(2) Both the methodology [RFC] uses in conducting its re-underwriting analyses
of the loans included in the Samples and the results of those re-underwriting
analyses;
(3) The methodology [RFC] ultimately uses for extrapolating the results of its
reunderwriting analysis for any Sample to the Population from which the
Sample was drawn;
(4) The weight a fact-finder should afford to any purported breach rate for any
Sample or Population, including due to the margin of error associated with
extrapolating the Breach Rate to the Population; and
(5) Any opinions expressed by any of [RFC’s] experts to the extent those opinions
(a) are not expressly set forth in the Sampling Protocol in Dr. Snow’s
declaration, (b) differ in any way from the statements set forth in the Sampling
Protocol in Dr. Snow’s declaration, or (c) are based on information not set
forth in the Sampling Protocol in Dr. Snow’s declaration.
In re Residential Capital, LLC, No. 14-ap-7900-mg, Doc. No. 66 ¶ 14 (Bankr. S.D.N.Y. Feb.
10, 2015).
10
This Court concludes that early decisions on the sampling issues will streamline the
administration of these very complex cases, and therefore grants the sampling motion in part and
denies the motion in part without prejudice. Accordingly, this Court grants the sampling motion
only to the extent that RFC seeks the limited ruling that: (1) Dr. Snow is a qualified expert
witness with respect to the selection and construction of RFC’s proposed samples, and the
extrapolation of a breach rate from those samples to the populations from which they were
drawn; (2) subject to the full reservation of the Defendants’ rights as set forth in this Order, the
sampling protocol set forth in Dr. Snow’s declaration is scientifically valid and admissible in
these cases for the purpose of identifying a random sample of loans from the population of loans
at issue in each case; and (3) Defendants have agreed they will not challenge that a 150-loan
sample is sufficiently large to identify a breach rate with a margin of error at the 95% confidence
level of no more than +/- 8 percentage points for binary questions. To the extent RFC seeks any
further early Daubert or admissibility ruling, as it did when it initially filed its sampling motion,
the sampling motion is denied without prejudice.
The Court will add a statistical sampling disclosure schedule to the case management
order in this case. The parties are ordered to meet and confer and to submit to the Court two (2)
days before the status conference on April 24, 2015, their proposal(s) for a statistical sampling
disclosure schedule including the nature of timing of disclosures and objections. The parties
should be guided in this effort by the discovery schedule already in effect, and by the procedures
adopted in the parallel litigation before Judge Glenn in the bankruptcy proceedings in the
Southern District of New York. (See Doc. No. 257-3, Feb. 10, 2015 Order in In re Residential
Capital, LLC, No. 14-ap-7900-mg, Doc. No. 66 ¶¶ 16-17.)
11
THEREFORE, IT IS HEREBY ORDERED THAT:
Plaintiffs’ Motion to Approve Use of Statistical Sampling [Doc. No. 154] is GRANTED
IN PART and DENIED IN PART WITHOUT PREJUDICE, as set forth herein.
Date: April 16, 2015
s/Susan Richard Nelson
SUSAN RICHARD NELSON
United States District Judge
12
Disclaimer: Justia Dockets & Filings provides public litigation records from the federal appellate and district courts. These filings and docket sheets should not be considered findings of fact or liability, nor do they necessarily reflect the view of Justia.
Why Is My Information Online?