Federal Deposit Insurance Corporation v. Merrill Lynch Pierce Fenner & Smith, Inc. et al
Filing
223
ORDER DENYING Defendant's 192 Sealed Motion to exclude the testimony of Normal Miller. ORDER DENYING Defendant's 194 Sealed Motion to exclude the testimony of Dawn Molitor-Gennrich. Signed by Judge Xavier Rodriguez. (lt)
IN THE UNITED STATES DISTRICT COURT
FOR THE WESTERN DISTRICT OF TEXAS
AUSTIN DIVISION
FEDERAL DEPOSIT INSURANCE
CORPORATION, as receiver for
GUARANTY BANK,
§
§
§
§
§
§
§
§
§
§
§
Plaintiff,
v.
RBS SECURITIES, INC.,
Defendant.
CIVIL NO. 1-14-CV-126-XR
ORDER ON MOTIONS TO EXCLUDE EXPERT TESTIMONY
On this date, the Court considered the remaining motions to exclude expert testimony
pending in this case. 1 A previous order decided the motion of Plaintiff Federal Deposit
Insurance Company (“FDIC”) to exclude John Contino and the motions of Defendant RBS
Securities, Inc. (“RBS”) to exclude John Finnerty and Stephen Ryan. That order set out the
relevant factual and procedural background in this case, which the Court does not restate
here. Now, the Court will decide the remaining motions concerning Norman Miller (docket
no. 192) and Dawn Molitor-Gennrich (docket no. 194).
LEGAL STANDARD
Rule 702 of the Federal Rules of Evidence provides for the admissibility of expert
testimony if it will “help the trier of fact to understand the evidence or to determine a fact in
1
The case is related to FDIC v. Deutsche Bank Securities, Inc., No. 1-14-129-XR. The parties—who share
experts with that case—filed identical motions, responses, and replies in the two cases. Indeed, RBS and
Deutsche Bank filed their documents jointly. The Court’s analysis of these motions in RBS will thus be
identical to its analysis of the motions in Deutsche Bank. The Court, then, in following the practice of Judge
Sparks, who formerly handled this case prior to its transfer, will docket an order in Deutsche Bank that
adopts the reasoning presented below. Any reference to “RBS” below can thus be read as referring to
“Deutsche Bank” or “defendants,” unless otherwise indicated.
1
issue.” FED. R. EVID. 702. Additionally, the testimony must be “based on sufficient facts or
data” and be “the product of reliable principles and methods” that the expert has “reliably
applied” to the facts of the case at hand. Id.
As a preliminary matter, the Court must determine whether the proffered witness
qualifies as an expert. “Before a district court may allow a witness to testify as an expert, it
must be assured that the proffered witness is qualified to testify by virtue of his ‘knowledge,
skill, experience, training, or education.’” United States v. Cooks, 589 F.3d 173, 179 (5th
Cir. 2009) (quoting FED. R. EVID. 702). Generally, if there is some reasonable indication of
qualifications, the court may admit the expert's testimony, and then the expert's
qualifications become an issue for the trier of fact, rather than for the court. Rushing v.
Kansas City S. Ry. Co., 185 F.3d 496, 507 (5th Cir. 1999).
If the expert is qualified, then the Supreme Court’s decision in Daubert v. Merrell
Dow Pharmaceuticals, Inc., 509 U.S. 579 (1993) provides the analytical framework for
determining the admissibility of expert testimony. Daubert requires the district courts to act
as “gatekeepers” to ensure expert testimony meets Rule 702’s standards. Id. at 589. This
entails ensuring “that an expert's testimony both rests on a reliable foundation and is relevant
to the task at hand.” Id. at 597; Pipitone v. Biomatrix, Inc., 288 F.3d 239, 244 (5th Cir. 2002)
(“In short, expert testimony is admissible only if it is both relevant and reliable.”).
The reliability inquiry entails a preliminary assessment of whether the reasoning or
methodology underlying the testimony is scientifically valid and can be properly applied to
the facts in issue.
Id. at 592–93.
In Daubert, the Supreme Court enumerated five
nonexclusive factors to consider when assessing whether the methodology upon which an
expert rests his opinion is reliable. These factors are: (1) whether the expert's theory can be
or has been tested, (2) whether the theory has been subject to peer review and publication,
2
(3) the known or potential rate of error of a technique or theory when applied, (4) the
existence and maintenance of standards and controls, and (5) the degree to which the
technique or theory has been generally accepted in the scientific community. Id. at 593–94;
Burleson v. Tex. Dep’t of Criminal Justice, 393 F.3d 577, 584 (5th Cir. 2004).
The test for determining reliability is flexible and can adapt to the particular
circumstances underlying the testimony at issue. Kumho Tire Co., Ltd. v. Carmichael, 526
U.S. 137 (1999). The point of this inquiry “is to make certain that an expert, whether basing
testimony upon professional studies or personal experience, employs in the courtroom the
same level of intellectual rigor that characterizes the practice of an expert in the relevant
field.” Id.
In applying the Daubert test, the proponent of expert testimony has the burden to
prove by a preponderance of the evidence that evidence is reliable (not that it is correct).
Moore v. Ashland Chem. Inc., 151 F.3d 269, 276 (5th Cir. 1998). Expert testimony must be
reliable “at each and every step” because “[t]he reliability inquiry applies to all aspects of an
expert's testimony: the methodology, the facts underlying the expert's opinion, the link
between the facts and the conclusion, et alia.” Knight v. Kirby Inland Marine Inc., 482 F.3d
347, 355 (5th Cir. 2007) (quoting Heller v. Shaw Indus, Inc., 167 F.3d 146, 155 (3d Cir.
1999)). “[I]n determining the admissibility of expert testimony, the district court should
approach its task ‘with proper deference to the jury's role as the arbiter of disputes between
conflicting opinions.” United States v. 14.38 Acres of Land, More or Less Sit. in Leflore
County, Miss., 80 F.3d 1074, 1077 (5th Cir. 1996) (quoting Viterbo v. Dow Chemical Co.,
826 F.2d 420, 422 (5th Cir. 1987)).
3
DISCUSSION
Under the provision of the Texas Securities Act at issue here, “[a] person who . . .
sells a security . . . by means of an untrue statement of a material fact or an omission to state
a material fact . . . is liable to the person buying the security from him, who may sue either
at law or in equity for rescission, or for damages if the buyer no longer owns the security.”
TEX. REV. CIV. STAT. ART. 581-33(A)(2). A seller is not liable, however, if it can prove
“either (a) the buyer knew of the untruth or omission or (b) he (the offeror or seller) did not
know, and in the exercise of reasonable care could not have known, of the untruth or
omission.” Id.
1. RBS’s Motion to Exclude Norman Miller’s Testimony
RBS moves to exclude the testimony of Norman Miller (docket no. 192).
a. Qualifications
First, this Court must determine if Miller is qualified to testify on the matters at hand.
Cooks, 589 F.3d at 179. A party who moves to exclude expert testimony on qualification
grounds must show that the expert does not possess a higher degree of knowledge, skill,
experience, or education than an ordinary person. See McCullock v. H.B. Fuller Co., 61 F.3d
1038, 1043 (2d Cir. 1995). RBS does not challenge Miller’s qualifications, but the Court, in
its gatekeeping function, will assess his qualifications and experience.
Miller is a Professor of Real Estate at the University of San Diego, and before that
he taught at the University of Cincinnati. Docket no. 192-2 at 6. He earned a Ph.D. from
Ohio State University in 1977. Id. He is a past president of the American Real Estate Society
and is on the editorial boards of several journals. Id. He has consulted on mortgage risk
analysis and automated valuation models (“AVMs”) in the private sector, and housing
market analysis has been the focus of many of his academic publications. Id. at 6-7. His
4
research expertise in pricing models has led firms to use his methods in the private sector,
and as advisor to these firms Miller has “applied various property valuation methods and
modeling techniques, including hedonic modeling and appraisal emulation, to develop
automated valuation tools for use by the firms’ clients.” Id. at 7. His “work on AVMs focuses
on methodologies to improve the accuracy of the models by identifying specific influences
on value, and developing and testing modules that improve accuracy.” Id. He has been an
expert witness in RMBS cases, and on valuation disputes that predate these cases. Id.
Having reviewed Miller’s report, the Court is satisfied that Miller has the necessary
expertise to testify on its contents, to the extent they are reliable and relevant.
b. Reliability and Relevance of the Opinions
Next, the Court considers the reliability and relevance of Miller’s opinions.
i. Miller’s Report
The FDIC retained Miller to “provide an expert opinion relating to the selection and
use of a retrospective automated valuation model (AVM) to calculate the weighted average
loan-to-value ratios (LTVs) of the pools of loans that provide cash flows to the certificates
Guaranty Bank purchased[.]” Id. at 5. This testimony pertains to the FDIC’s contention that
RBS’s prospectus supplements materially misstated the LTVs of the loans underlying the
Certificates sold to Guaranty Bank.
First, Miller opines that a retrospective AVM is an appropriate, reliable way to
estimate the values of properties as of 2004 and 2005 in calculating the weighted average
LTVs of the pools. Id. at 5. He explains that AVMs estimate the value of a given property
as of a given date. Id. at 9. When valuing a property, an appraisal emulation model of AVM,
the type Miller used here, identifies comparable properties (ranked by similarity to the
subject property), then accounts for differences in features and size and historical price
5
trends in the area. Id. An AVM’s value estimate includes a confidence score that indicates
its accuracy. Id. at 11. Miller states that “[b]ecause of their efficiency and reliability, AVMs
were frequently used by lenders and investment banks during the time relevant to this
litigation, and continue to be used in the real estate industry widely today.” Id.
Miller states that “manual appraisals traditionally have done a better job than AVMs
of accounting for unique features or adjusting for the condition of a subject property,” but
“so long as the AVM is run over a reasonable number of loans, the positive and negative
errors tend to cancel out.” Id. at 17. Miller argues AVMs compare well to manual appraisals
because AVMs use precise statistical methods to adjust sales prices of comparable properties
when accounting for differences in size and features, AVMs are not prone to bias, AVMs’
confidence metrics allow users to quantify the risk of error in an estimate, and AVMs have
become more accurate over the years with better data filtering, especially among “leading
venders like DataQuick.” Id.
Second, Miller opines that the retro appraisal emulation AVM developed by
DataQuick Information Systems, Inc. (“DataQuick”) is a reliable AVM for this case. Id. at
5. He selected it based on his knowledge of DataQuick’s reputation in the AVM industry,
his “detailed discussions with DataQuick about its specific model,” and his knowledge of
appraisal emulation models generally. Miller opines that this AVM “is the best available tool
because it follows the same methodology as a human appraiser: it selects the best available
comparables, makes adjustments for features and time, and weights the comparables to come
up with an estimate of market value.” Id. at 12. Miller also states he reviewed a DataQuick
white paper that describes the AVM’s methodology. Id. at 12-13. He states he “interviewed
the AVM developers at DataQuick and was satisfied from my discussions with [lead AVM
developer Dr. Gordon] Crawford and his colleagues that the DataQuick Appraisal Emulation
6
AVM follows a reliable methodology and would produce accurate results when run on pools
of loans.” Id. at 13. He confirmed the reliability of the values produced by the DataQuick
AVM with an analysis of the default rates of loans relative to their LTVs based on AVM
values.
Third, Miller opines that the appraisal values on which the original weighted average
LTVs were based are biased in favor of overvaluation. Id. at 5. He bases this opinion on his
review of LTVs based on appraisal values, LTVs based on AVM values, and the
corresponding default rates of the loans. Applied here, the AVM returned values for 9,063
of 10,567 properties, with an average confidence score of 83. Id. at 16. Miller removed 7
properties for which the entries in the origination date field in the loan tapes may have been
incorrect, and the FDIC’s statistical expert removed 184 loans as outliers. Id. The report uses
the remaining 8,879 loans. Comparing the weighted average LTVs as recalculated by the
AVM to the original weighted average LTVs included in the prospectus supplements, Miller
concludes that the supplements understated the LTVs. Id. at 19. He argues this shows
“investors in each of these securitizations were looking at significantly less real equity to
protect against losses in foreclosure than the LTVs in the loan tapes suggest.” Id. at 20. Thus,
“[t]he risks of losses in foreclosure, based on such differences in value estimation, were
significantly higher than those disclosed on the basis of original appraisals.” Id. He also
argues his opinion is supported by empirical analysis of large samples of mortgage loans in
recent academic literature. Id.
Thus, fourth, Miller opines that the weighted average LTVs based on AVM values
are more accurate estimates of the borrowers’ equity in the properties that back the loans at
issue. Miller examined whether the rate of default was higher for the loans at issue in this
case with original LTVs below 80 percent but AVM-based LTVs above 80 percent. Id. at
7
21. He chose this threshold because, he contends, lendors and investors consider loans with
LTVs above 80 percent to be the riskiest because lenders are unlikely to recover any equity
in the event of foreclosure. Id. He found a 32.71 percent default rate on the loans with an
original LTV below 80 percent and an AVM-based LTV above 80 percent, as compared to
a 19.57 percent default rate on the loans with both original and AVM-based LTVs below 80
percent. Id. at 22.
ii. Discussion
RBS seeks to exclude Miller’s testimony as unreliable for several reasons. Docket
no. 192.
1. Challenge to Retrospective Use of DataQuick AVM
First, RBS argues Miller used the DataQuick AVM as his “sole basis for his opinion
that the value opinions provided by the professional appraisers, who conducted on-site
appraisals for the properties associates with the loans underlying the Certificates, were
overvalued.” Id. at 10. RBS contends this reliance renders his testimony unreliable.
RBS’s principal support for this contention is a 2016 deposition of Stanley Wu, Vice
President of Data Science and Architecture for CoreLogic, the company that owns the
DataQuick AVM. Id. at 9. Wu testified that “[p]rofessionals in the real estate field should
not—and I believe, do not—rely solely on CoreLogic or other AVMs to make reliable
determinations of the reasonableness of value opinions offered by licensed certified
appraisers, whether individually or collectively.” Id. Wu further testified that CoreLogic’s
“AVMs are not used to determine whether an appraiser actually inflated or deflated an
opinion of value . . . . because there’s no way to discern from AVMs the cause of any
difference between an AVM’s point estimate and the opinion of value” in a licensed
appraisal. Id. Wu stated that “even where a difference is identified between AVM value
8
ranges and appraisals’ opinions of value, no professional would purport to draw conclusions
on opinions of value contained in professional performed appraisals.” Id.
In response, the FDIC deflects Wu’s testimony, pointing to contradictory statements
made in marketing materials that stated, for example, “[a] comparison between our
retroactive AVM appraisals of properties and the originally appraised values can detect
disparities that call into question the earlier appraisals and indicate possible loan-to-value
misrepresentations.” Docket no. 207 at 16.
In addition to its purported conflict with CoreLogic’s recommended use, RBS argues
Miller’s use of a single AVM conflicts with industry standards (citing a report from RBS
expert Lee Kennedy) and regulatory standards (citing the Interagency Appraisal and
Evaluations Guidelines, which state an AVM user “should not rely solely on validation
representations provided by an AVM vendor”). Id. at 12. Further, RBS cites an RMBS case
that rejected use of an AVM because it could not account for “important considerations in
valuing a residential property” like whether a home has a view of the ocean or a parking lot.
See U.S. Bank, Nat’l Ass’n v. UBS Real Estate Sec. Inc., 205 F. Supp. 3d 386, 434 (S.D.N.Y.
2016).
The FDIC distinguishes this UBS case, which considered the reliability of the AVM
“with respect to an individual property, as distinguished from a ‘mass basis,’” and in which
the plaintiff had to prove that the values of specific, individual properties were inflated.
Docket no. 207 at 17-18. The FDIC argues that here, by contrast, Miller used the AVM to
estimate average LTVs of pools of loans, not second-guess an appraiser on any individual
property. Id. at 18. Further, Miller explains that the use of multiple AVMs—called a
“cascade”—“may be appropriate when the user wishes to value individual properties in
9
several specific markets,” but “the high accuracy of the results when the AVM is used to
determine the average LTV of a pool of loans obviates the need for a cascade.” Id.
RBS also points to Wu’s testimony in arguing that, even if the DataQuick AVM
could reliably draw conclusions about appraisal values, it cannot do so retrospectively, at
least not to the extent Miller attempts. Wu testified that he was not aware of any tests of the
DataQuick AVM looking five or ten years in the past, and an article by CoreLogic’s Vice
President of Collateral Solutions states the retrospective testing of the AVM “only went back
two years, so further testing would be needed to determine how far back is too far back for
retrospective AVMs to be useful.” Docket no. 192 at 10-11.
The FDIC counters this argument by citing several cases that have accepted
testimony based on retrospective AVMs. Docket no. 207 at 8. Further, it argues Miller and
other academics rely on retrospective AVMs in their research to “determine values of
properties many years in the past, including to evaluate the presence of appraisal fraud and
appraisal inflation during the build-up to the financial crisis.” Id. (citing docket no. 201-2 at
6 (listing academic articles that rely on output of retrospective AVMs)).
Finally, RBS argues the DataQuick AVM systematically underestimates the value of
the properties as compared to actual sale prices. RBS expert Jerry Hausman performed a bias
test on the DataQuick AVM, comparing the AVM values for the relevant purchase
transactions to the actual purchase prices for those properties. Docket no. 192 at 14.
Hausman concluded the average AVM values is below the average purchase price for all
loan groups. Id. at 15. RBS contends this is to be expected, given Miller’s concession that
“AVMs tend to produce less accurate values in rapidly fluctuating, highly-dispersed, highlypriced dispersed markets.” Id. As to these arguments, while Miller refutes them in his
declaration, the FDIC contends they go only to weight and should be resolved at trial.
10
Here, the Court finds unavailing RBS’s above arguments as to the reliability of
AVMs generally, Miller’s chosen DataQuick model specifically, or this model’s
retrospective use in the context of this case. As have other courts, this Court will allow the
retrospective use of a commercial AVM in assessing LTVs of large groups of mortgage
loans. The FDIC lists several cases that have allowed testimony based on AVMs.2
The Court finds one case particularly persuasive. See docket no. 207-4, Order
Denying Defendants’ Motion to Exclude the Testimony of Dr. Norm G. Miller, Fed. Home
Loan Bank of Seattle v. Banc of Am. Secs. LLC, Nos. 09-2-46319-1 et al. (Sup. Ct. King Cty.,
Wash., Mar. 16, 2016) (“FHLB Seattle”).3 Applying a rule modeled on Rule 702, the FHLB
Seattle court rejected similar challenges to similar testimony by the same expert considered
here, Norman Miller. The court stated that
[t]he purpose for which the DataQuick AVM is being offered, to value large
pools of properties, has been generally accepted by the courts, the real estate
industry, [and] academics . . . for years. It is not, as argued by Defendants,
being offered for the purpose of ‘overturning the value opinions of appraiser.’
This evidence is generally accepted in the relevant community.
Id. at 5-6.
RBS raises grounds for distinguishing the FDIC’s cited cases, including by noting
that the experts in some of those cases used a different commercial AVM or developed their
2
Docket no. 207 at 8 (citing Mass. Mut. Life Ins. Co. v. DB Structured Prods., Inc., No. 11-30039-MGM,
2015 WL 2130060, at *9 (D. Mass. May 7, 2015); Fed. Hous. Fin. Agency v. Nomura Holding Am., Inc., No.
11-CV-6201, 2015 WL 353929, at *4 (S.D.N.Y. Jan. 28, 2015); Fed. Home Loan Bank of Seattle v. Banc of
Am. Secs. LLC, Nos. 09-2-46319-1 et al. (Sup. Ct. King Cty., Wash., Mar. 16, 2016); FDIC as Receiver for
Franklin Bank v. Morgan Stanley & Co. LLC, No. 201167305 (Dist. Ct. Harris Cty., Tex., Feb. 25, 2015);
CMFG Life Ins. Co. v. RBS Secs. Inc., No. 12-CV-037-WMC, 2014 WL 3696233 (W.D. Wis. July 23, 2014),
rev’d in part on other grounds, No. 14-2904, 2015 WL 4978996 (7th Cir. Aug. 21, 2015); Assured Guar.
Mun. Corp. v. Flagstar Bank, FSB, 920 F. Supp. 2d 475, 505 (S.D.N.Y. 2013)).
3
The Court also takes note of a Texas state court that denied a challenge to AVM testimony offered by
Miller. See docket no. 207-5, Order on Defendants’ Motion to Exclude Experts, FDIC as Receiver for
Franklin Bank v. Morgan Stanley & Co. LLC, No. 201167305 (Dist. Ct. Harris Cty., Tex., Feb. 25, 2015).
This order contains no analysis, however, so it is useful here only to the extent it indicates other courts have
found Miller’s testimony acceptable.
11
own. Docket no. 216-1 at 9. The cases that use the same AVM Miller uses here, RBS argues,
did not have the benefit of Wu’s testimony casting doubt on testimony like Miller’s. Still,
the FDIC points out that Wu adopted verbatim the statements of an earlier affidavit by a
CoreLogic vice president, Jacqueline Doty, and Doty’s statements were available to and
rejected by the FHLB Seattle court. Id. at 16-17 (citing FHLB Seattle, docket no. 207-4 (“To
the extent [the issues raised in Ms. Doty’s affidavit relate to Dr. Miller’s use,] the court finds
the position of Dr. Miller more persuasive.”)). Given the conflicting statements from
CoreLogic in the record, the Court does not find Wu’s testimony warrants excluding Miller’s
testimony. More broadly, RBS’s distinctions of the FDIC’s cited cases do not affect the
Court’s conclusion, supported by the weight of case law, that AVM testimony like Miller’s
clears the threshold of reliability for Daubert purposes.
Thus, RBS’s view that Miller should have used more AVMs (or that it should not
have used this one retrospectively) and its arguments regarding Wu’s testimony will go to
weight, not admissibility. Having decided these more general attacks on the reliability of
AVMs used as Miller does here, the Court will turn to RBS’s arguments that Miller’s
understanding of the AVM is inadequate.
2. Challenges to Miller’s Testing and Knowledge of the AVM
RBS advances several arguments as to Miller’s evaluation, testing, and knowledge
of the AVM and its methodology. First, RBS argues Miller’s results cannot be replicated.
RBS expert Lee Kennedy placed an order with DataQuick using the properties, customized
settings, and retrospective dates used by Miller. Docket no. 192 at 13. RBS contends
Kennedy’s results deviated from Miller’s significantly—for example, Kennedy was
provided 9,041 hits on the 10,567 properties, while Miller was provided 9,065. Id. The FDIC
contends Kennedy received almost identical results because “[t]he average AVM value—
12
the relevant metric here—is $526,738 from Dr. Miller’s run and $526,814 from Mr.
Kennedy’s run, a difference of $77, or 0.01%.” Docket no. 207 at 20-21.
Next, RBS argues Miller has “only a superficial understanding” of the DataQuick
AVM, as his knowledge of it comes from “a generic marketing document and his vague
recollection, unsupported by contemporaneous notes, of a few brief conversations with
former DataQuick personnel several years ago.” Id. at 6. RBS argues Miller’s conversations
with DataQuick developers were only three phone calls conducted years ago. Id. at 16. RBS
contends he “simply relied on a ‘screen shot’ of purported internal and external testing
results from several years ago provided by Dr. Crawford,” and “he did not even look at the
external testing results for the DataQuick AVM he decided to use here[.]” Id.
The FDIC argues Miller applies his expertise and used due diligence in selecting the
AVM. The FDIC points to the acceptance of Miller by the FHLB Seattle court, which found
[Miller’s] declaration and reports demonstrate he has substantial expertise in
AVMs and methodologies that underlie them, including the appraisal
emulation process, and inputs and data on which the DataQuick AVM relies.
This comes from his own work, in addition to interviews with DataQuick
personnel and investigation into its methodologies.
Docket no. 207 at 10. Contesting RBS’s portrayal of Miller’s due diligence, the FDIC states
he interviewed the AVM’s head developer for several hours, which is a “substantial length
of time for experts to discuss a subject they already know well.” Id.
Finally, RBS argues Miller does not have the “information needed to competently
evaluate the DataQuick AVM, including essential information about its inputs, outputs and
methodology.” Id. at 6. He “did not review any of the code or algorithms used to program
the DataQuick AVM . . . and thus cannot speak with any certainty about the specific process
used by DataQuick to generate the AVM’s opinions of value.” Id. at 17. Because of this,
RBS argues the DataQuick AVM is an “impermissible black box,” and Miller’s lack of
13
knowledge concerning its inputs makes it impossible for RBS to effectively cross-examine
him. Id.
The FDIC, in response, argues courts regularly admit opinions based on proprietary
software without requiring the testifying expert to review the underlying code, so long as the
program was commonly used in the expert’s field, and the FDIC argues AVMs from
commercial vendors are widely used in the real estate industry and academia. Docket no.
207 at 12-13. Further, several of the cases FDIC cites allowed commercial AVMs. See FHLB
Seattle, docket no. 207-4 at 7-8; Franklin, docket no. 207-5 at 1; CMFG, 2014 WL 3696233,
at *14; Assured Guar., 920 F. Supp. 2d at 505.
Having reviewed Miller’s report and his declaration in support of the FDIC’s
response, the Court is satisfied that Miller adequately vetted the DataQuick AVM, and any
perceived inadequacies in his evaluation can be explored on cross-examination. This is true
also of RBS’s concerns as to replicability. Finally, given Miller’s experience with AVMs,
his expertise in their methodologies, and the routine acceptance of commercial AVMs by
other courts, the Court does not deem it necessary that Miller access the underlying source
code of the AVM used. As the FHLB Seattle court stated,
Miller is able to determine if there is a code error based upon output of the
AVMs. Nor is it necessary, or even possible to review ‘an algorithm,’ as there
is no single mathematical code to support the model. There are potentially
millions of mathematical combinations for making adjustments in the
appraisal emulation model. What is necessary to test the model was in fact
performed, as extensively described in the report[] and declaration[] of Miller
....
FHLB Seattle, docket no. 207-4 at 6.
Thus, RBS’s motion to exclude Norman Miller’s testimony is DENIED.
14
2. RBS’s Motion to Exclude Dawn Molitor-Gennrich’s Testimony
Next, the Court turns to RBS’s motion to exclude Dawn Molitor-Gennrich (docket
no. 194).
a. Qualifications
First, this Court must determine if Molitor-Gennrich is qualified to testify on the
matters at hand. Cooks, F589 F.3d at 179. A party who moves to exclude expert testimony
on qualification grounds must show that the expert does not possess a higher degree of
knowledge, skill, experience, or education than an ordinary person. See McCullock, 61 F.3d
at 1043. RBS does not appear to challenge Molitor-Gennrich’s qualifications as an expert,
but the Court will assess Molitor-Gennrich’s qualifications and experience.
Molitor-Gennrich is a private consultant on matters of real property appraisals and
appraisal services. She is a member of the Appraisal Institute (and was on its Board of
Directors) and holds designations as a member who can provide reviews of residential
appraisals and opinions of value, evaluations, reviews, consulting, and advice regarding
investment decisions. Docket no. 187-3 at 7. She has, since 2009, “provided my expertise in
the Uniform Standards of Professional Appraisal Practice (USPAP or the Uniform
Standards) and the Appraisal Institute’s Code of Professional Ethics as a peer review
screener for the Appraisal Institute’s Professional Practice area.” Id.
Between 2002 and 2007, she served on the Appraisal Foundation’s Appraisal
Standards Board, which develops, interprets, and amends the USPAP. She is an Appraiser
Qualifications Board-certified USPAP instructor. She previously worked for Union Bank
and for the Bank of California in different appraisal-related capacities. Id. at 8. After a
previous firm she co-founded ceased operations, she founded Dawn Molitor-Gennrich
Consulting, LLC, which “provides consulting and litigation support services in the areas of
15
real property appraisal and appraisal review, including real estate valuations standards and
guidelines, such as USPAP[.]” Id.
Having reviewed Molitor-Gennrich’s report, the Court is satisfied that she has the
necessary expertise to testify on its contents, to the extent they are reliable and relevant.
b. Molitor-Gennrich’s Report
Molitor-Gennrich was asked to evaluate the credibility, under USPAP, of Appraisal
Reports relating to 666 of the underlying residential mortgage loans selected by another
expert, Charles Cowan. Docket no. 187-3 at 22. This testimony pertains to the FDIC’s claim
that RBS’s prospectus supplements materially misstated that the appraisals related to the
loans underlying the RMBS at issue complied with USPAP.
Molitor-Gennrich states USPAP divides the appraisal process into development (the
minimum practice requirements, which are set out in USPAP Standard 1) and reporting
(Standard 2). Id. at 12. Standard 3 states the requirements for conducting an appraisal review.
Id. Also relevant are the Ethics Rule (which requires that the appraiser act with the right
intention) and the Competency Rule (which requires both that an appraiser have the
knowledge to appraise a property and apply appropriate analytical methods in doing so). Id.
at 13-15.
For each of the 666 loans, Molitor-Gennrich reviewed, if available, the Appraisal
Report and other supplemental documents from the loan file, including the agreements of
sale for the subject properties, contemporaneous reviews of the Appraisal Reports, and any
workfiles that may have been included with the Appraisal Reports. Id. at 22-23. She
developed a checklist form (the “Form”) to collect information for each Appraisal Report,
the intent of which was to see whether the Appraisal Report differed from various data
sources available at the time the report was created, including Multiple Listing Services,
16
online local public records data, and aggregate public data sources (collectively, “Appraisal
Materials”). Id. at 23. She hired a firm to complete the Form, trained the firm’s appraisers to
use the Form, and trained a separate team of appraisers to compare the results of the Form
to the Appraisal Materials. Id. This latter team then screened the Forms and passed on to
Molitor-Gennrich a subset that excluded those Forms with factual data discrepancies with
the Appraisal Report. Id.
For this subset (375 of the 666 loans), she conducted a Standard 3 compliance review.
Id. at 24. This involved a review of the Appraisal Report, Appraisal Materials, the Form, and
any Supplemental Data. Id. This review included “viewing maps and aerial imagery of each
subject property and the surrounding area from around the time period when the associated
appraisal was conducted.” Id. She recorded her opinions as to the credibility of the Appraisal
Reports and compliance with the Competency Rule in another form she created, the USPAP
Questionnaire. Id.
She states that, while minimal discrepancies did not trigger non-compliance in her
review,
an appraisal report is not credible if the factual data presented is determined
to be inaccurate or cannot be confirmed to be accurate from reliable data
sources; the analysis of the data is not clear; the valuation approaches are not
appropriate or applied appropriately for the assignment; or the analysis
required throughout the appraisal report and the reconciliation of conclusions
does not tie the appraisal report together in a way that intended users can
understand the rationale for the opinions and conclusions for the intended
use.
Id. After review, she found that 365 of the 375 Appraisal Reports she reviewed were not
credible. Id. She deemed a report not credible if it failed to comply with two of more of the
benchmark rules in Standard 1: “[b]e aware of, understand, and correctly employ recognized
methods/techniques necessary to produce a credible appraisal” (1-1(a)); “[d]o not commit a
17
substantial error of omission or commission that significantly affects an appraisal” (1-1(b));
“[d]o not render appraisal services in a careless or negligent manner, such as making a series
of errors that, in aggregate, affect appraisal credibility” (1-1(c)). Id. at 26. The bulk of
Molitor-Gennrich’s report is comprised of illustrative examples of the ways specific
appraisals fall short.
c. Reliability
RBS argues Molitor-Gennrich’s opinions are unreliable because she improperly
relied on Appraisal Reports (without reviewing the appraisers’ workfiles or interviewing
appraisers) and on third-party data.
First, RBS argues Molitor-Gennrich’s reliance on the Appraisal Reports, prepared to
report appraisal findings to clients, is improper because they are “merely summaries and did
not purport to contain all information used in producing the report.” Docket no. 194 at 12.
Because USPAP requires appraisers to prepare “workfiles” with information “necessary to
support the appraiser’s opinions and conclusions and to show compliance with . . . applicable
Standards,” RBS contends Molitor-Gennrich should have requested and reviewed these
workfiles. Id. Alternatively, RBS argues she should have interviewed the appraisers to
determine whether information that is not otherwise required in an Appraisal Report was
actually missing from the appraisers’ analysis. Id. In RBS’s view, to conclude that the
appraisals fell short of USPAP, “[t]he fact that details were not addressed in Summary
Appraisal Reports is woefully insufficient given that, by definition, such summary reports
were not intended to contain the detailed analyses Ms. Molitor-Gennrich claims to be
missing.” Id. at 13.
In response, the FDIC argues appraisal reviews do not require review of workfiles or
appraiser interviews because USPAP states “[t]he subject of an appraisal review assignment
18
may be all or part of a report, workfile, or a combination of these.” Docket no. 208 at 11.
The FDIC notes that, according to USPAP, “a summary Appraisal Report must include
sufficient information to indicate that the appraiser complied with the requirements of
STANDARD 1” and must “summarize the information analyzed, the appraisal procedures
followed, and the reasoning that supports the analyses, opinions, and conclusions.” Id. at 12.
The FDIC states no court has considered whether appraiser interviews are necessary, but the
only court that has considered whether workfiles are necessary rejected that argument. Id. at
13 (citing Mass. Mutual Life Ins. Co. v. DB Structured Prods., Inc., No. 11-30039-MGM,
2015 WL 2130060, at *11 (D. Mass. May 7, 2015)).
RBS distinguishes that case by arguing the expert in question, John Kilpatrick,
applied “a wholly different methodology” than does Molitor-Gennrich and the court does
not analyze the role of workfiles in that methodology. Docket no. 218-1 at 9. RBS is correct
that Kilpatrick used a different method, but that methodology did not include a review of the
workfiles (which the defendant in that case pointed to in attempting to exclude the testimony)
and yet the court allowed Kilpatrick to testify as to the appraisals’ credibility. Thus, based
on this and other arguments raised by the FDIC, the Court is persuaded that it is not
necessary, in clearing the Daubert reliability threshold, that Molitor-Gennrich have reviewed
appraisal workfiles or conducted appraiser interviews. These choices can be explored on
cross-examination.
Next, RBS argues Molitor-Gennrich’s opinions are unreliable because she relied on
unverified third-party data, including data from Google Earth, Multiple Listing Services, and
a housing price index. Docket no. 187-1 at 14-16. These arguments, however, go to weight,
not admissibility. See United States v. 14.38 Acres of Land, More or Less Situated in Leflore
Cty., State of Miss., 80 F.3d 1074, 1077 (5th Cir. 1996) (“As a general rule, questions relating
19
to the bases and sources of an expert's opinion affect the weight to be assigned that opinion
rather than its admissibility and should be left for the jury's consideration.”). Thus, the Court
finds Molitor-Gennrich’s testimony sufficiently reliable.
d. Relevance
Next, RBS argues Molitor-Gennrich’s testimony is not relevant. First, RBS argues
Molitor-Gennrich’s opinions are irrelevant because she admits she “did not make a judgment
as to whether the original appraisers honestly held the opinions of value that are stated in
their reports[.]” Docket no. 187-1 at 16. But it would have been improper for her to testify
as to the original appraisers’ state of mind, as multiple courts have concluded. See
Massachusetts Mut. Life Ins. Co., 2015 WL 2130060, at *13 (D. Mass. May 7, 2015) (“Dr.
Kilpatrick may not directly testify as to the original appraisers' state of mind, but he may
testify as to the appraisals’ ‘credibility.’”); Fed. Hous. Fin. Agency v. Nomura Holding Am.,
Inc., No. 11-CV-6201-DLC, 2015 WL 353929, at *6 (S.D.N.Y. Jan. 28, 2015) (“Evidence
of the extent to which the Nomura Appraisals conformed to industry standards is a distinct
inquiry from the issue of an individual appraiser's belief and whether it was honestly held.
While the former may provide circumstantial evidence regarding the latter, it is not
inadmissible on that score.”).
The Court adopts the following reasoning of the Nomura court:
The term “credibility” is a defined term in USPAP. It reflects an
objective assessment of an appraisal. It imposes on appraisers the obligation
to support an appraisal by evidence and logic. Thus, credible appraisals are
those that have complied with USPAP standards. Whether a particular
appraisal does so comply is a classic subject for expert testimony.
The defendants contend that Kilpatrick's testimony about the
credibility of appraisals is nothing more than an improper attempt to opine
on the honesty of appraisers. They are wrong. Evidence of the extent to which
the Nomura Appraisals conformed to industry standards is a distinct inquiry
from the issue of an individual appraiser's belief and whether it was honestly
20
held. While the former may provide circumstantial evidence regarding the
latter, it is not inadmissible on that score.
Nomura Holding Am., Inc., No. 11-CV-6201-DLC, 2015 WL 353929, at *6.4 In addition to
arguing that Molitor-Gennrich’s opinions that many appraisals were not credible represents
circumstantial evidence that the appraisers did not believe the values in their appraisals, the
FDIC also argues the opinions are relevant to the FDIC’s allegation that RBS’s statements
about LTV values were misleading. Docket no. 208 at 19. At summary judgment, Judge
Sparks noted this “could prove actionable even without evidence of subjective falsity.”
Docket no. 154 at 12. On either basis, the Court finds Molitor-Gennrich’s testimony relevant.
CONCLUSION
For the foregoing reasons, Defendant RBS’s motion to exclude the testimony of
Norman Miller (docket no. 192) is DENIED.
Further, Defendant RBS’s motion to exclude the testimony of Dawn MollitorGennrich (docket no. 194) is DENIED.
It is so ORDERED.
RBS argues the Nomura court’s reasoning should not apply because the expert there, Kilpatrick,
used a different methodology and the Nomura court based its holding on “other evidence” in the case. Docket
no. 218-1 at 11-12 (citing Nomura, 104 F. Supp. 3d 441, 508 (S.D.N.Y. 2015)). But the Nomura order RBS
cites sets out fact findings and legal conclusions following a bench trial. That order weighed whether, coupled
with other evidence, Kilpatrick’s opinions on the appraisals’ credibility was “sufficient circumstantial evidence
of bias to permit a determination that the appraisers produced appraisals that they knew did not accurately
describe the value of these properties.” The order the Court cites is an earlier ruling on the admissibility of
Kilpatrick’s testimony. Here, the Court does not decide whether Molitor-Gennrich’s testimony is sufficient
circumstantial evidence, only that the Court is satisfied, for relevance purposes, that it could constitute such
evidence, particularly when coupled with other evidence (as in Nomura).
4
21
SIGNED this 26th day of March, 2019.
XAVIER RODRIGUEZ
UNITED STATES DISTRICT JUDGE
22
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?