Assured Guar. Mun. Corp. v. Flagstar Bank

Full title: ASSURED GUARANTY MUNICIPAL CORP. f/k/a Financial Security Assurance Inc.…

Court: United States District Court, S.D. New York.

Date published: Feb 15, 2013

920 F. Supp. 2d 475 (S.D.N.Y. 2013)

Opinion

No. 11 Civ. 2375 JSR.

2013-02-15

ASSURED GUARANTY MUNICIPAL CORP. f/k/a Financial Security Assurance Inc., Plaintiff, v. FLAGSTAR BANK, FSB; Flagstar Capital Markets Corp.; and Flagstar ABS, LLC, Defendants.

Jacob W. Buchdahl, Arun Srinivas Subramanian, Susman Godfrey LLP, New York, NY, Jonathan Bridges, Joseph C. Portera, Warren T. Burns, Susman Godfrey LLP, Dallas, TX, Manmeet S. Walia, Stephen D. Susman, Susman Godfrey LLP, Houston, TX, for Plaintiff. Veronica Ellen Rendon, Daniel Marc Kuhn, Mary Elizabeth Sylvester, Monique Anne Gaylor, Stewart David Aaron, Arnold & Porter, LLP, Scott M. Berman, Friedman Kaplan Seiler & Adelman LLP, New York, NY, for Defendant.

JED S. RAKOFF

Jacob W. Buchdahl, Arun Srinivas Subramanian, Susman Godfrey LLP, New York, NY, Jonathan Bridges, Joseph C. Portera, Warren T. Burns, Susman Godfrey LLP, Dallas, TX, Manmeet S. Walia, Stephen D. Susman, Susman Godfrey LLP, Houston, TX, for Plaintiff. Veronica Ellen Rendon, Daniel Marc Kuhn, Mary Elizabeth Sylvester, Monique Anne Gaylor, Stewart David Aaron, Arnold & Porter, LLP, Scott M. Berman, Friedman Kaplan Seiler & Adelman LLP, New York, NY, for Defendant.

FINDINGS OF FACT, CONCLUSIONS OF LAW, AND ORDER

JED S. RAKOFF, District Judge.

Plaintiff Assured Guaranty Municipal Corporation (“Assured”) ¹ alleges that defendants Flagstar Bank, FSB; Flagstar Capital Markets Corporation; and Flagstar ABS, LLC (collectively, “Flagstar”) breached a series of contracts that provided financial guaranty insurance against certain defaults affecting two Flagstar securitizations backed by home equity loans. In particular, Assured alleges that the loans underlying the securities were either materially fraudulent or were the product of material underwriting defects, in breach of Flagstar's express representations and warranties. Because Flagstar refused to cure the defects or substitute eligible loans, Assured claims that it is entitled to be reimbursed for its payment of insurance claims that arose when many of the underlying loans defaulted. ²

1 At the time of the transactions at issue, plaintiff was known as Financial Security Assurance, Inc., or “FSA,” which Assured purchased in 2009. Trial Tr. 46:12–14. This Memorandum will refer to this entity as “Assured” throughout, except where it is particularly relevant to refer to it as “FSA.”

2 Assured also made claims relating to loan servicing, but withdrew those claims prior to trial.

Assured filed its complaint on April 7, 2011. Following discovery and motion practice that eliminated certain claims and remedies and clarified others, see Order, ECF No. 22 (July 7, 2011); Am. Mem., ECF No. 56 (Oct. 27, 2011); Order, ECF No. 84 (Feb. 29, 2012); Mem., ECF No. 100 (Sept. 25, 2012), the Court held a bench trial over twelve days between October 10 and November 12, 2012. The Court received into evidence 130 exhibits, including Flagstar's voluminous underwriting guidelines, as well as portions of the depositions of Matthew Roslin, Flagstar's former general counsel; Jean Garrick, Flagstar's head of quality control; George Stiehl, a former vice president in FSA's Residential Mortgage Group; David Beard, a former director of FSA's Corporate Finance Group; and David Williams, a former managing director in FSA's Residential Mortgage Group.³ The Court also heard live testimony from nine witnesses: Russell Brewer, Assured's Chief Surveillance Officer; Stanley Jursek, Flagstar Bank's Executive Vice President and Treasurer and the head of Flagstar Capital Markets Corp.; Marni Scott, Flagstar's Executive Vice President of Mortgage Credit Operations; Dr. Nelson Lipshutz, Assured's expert in statistical sampling; Rebecca Walzak, Assured's expert in mortgage underwriting and origination; Dr. Joseph Mason, Assured's damages expert; Jeffrey Nielsen, Flagstar's expert on the design and implementation of underwriting and due diligence reviews; Ann Rutledge, Flagstar's valuation expert; and John Griggs, Flagstar's expert in mortgage underwriting and origination. After the trial concluded, the Court received post-trial briefs from the parties on the issue of costs and expenses.

3 At the conclusion of the trial, the Court reserved ruling on certain objections made to various portions of the proffered deposition testimony. Although marked-up copies of deposition transcripts reflecting the Court's rulings on those depositions will in due course be filed, the findings of fact made herein reflect those rulings.

BASIC FACTS

Based on the evidence presented at trial, the facts stipulated to in the Joint Proposed Pretrial Order (“JPPO”), and the Court's assessment of the credibility and demeanor of the witnesses and the inferences reasonably to be drawn therefrom and from the evidence as a whole, the Court first makes the following findings of facts that set the stage for the central dispute in this case, the war of experts.

Flagstar Bank, a federally chartered savings bank located primarily in Michigan, provides residential mortgage loan origination services to its customers, including the origination of home equity lines of credit (HELOCs) and second-lien mortgages. JPPO ¶¶ A. 4, 7–8. Assured is a financial guaranty insurance company that provides bond insurance for, among other things, residential mortgage backed securities (“RMBS”). Trial Tr. 39:6–12.

Flagstar contracted with Assured in 2005 and 2006 to provide insurance on two securitizations of HELOCs, known as the “2005–1” and “2006–2” Trusts. JPPO ¶ A. 16. The 2005–1 Trust was collateralized by approximately 10,000 individual HELOC loans, totaling about $600 million, while the 2006–2 Trust was backed by over 5,000 HELOC loans, totaling about $300 million. JPPO ¶¶ A. 11; Trial Tr. 267:25–268:8; Pl. Ex. 76 at 75. All of the loans were either originated by Flagstar through its home lending group or purchased by Flagstar from mortgage loan brokers and correspondents throughout the United States. Either way, Flagstar represented to Assured that all of the loans were, at the time of origination, in compliance with Flagstar's underwriting guidelines. JPPO ¶¶ A. 14–15.

Investors in such trusts, known as bondholders, receive both monthly interest payments and an eventual return of principal. As a bond insurer, Assured guarantees timely payment of interest and principal to such bondholders, which raises the credit rating on the securitized loans and thus makes them more salable or salable at lower interest rates. Roslin Dep. Tr. 194:12–195:21. Here, Assured's participation in the 2005–1 and 2006–2 transactions raised the rating of the securitizations from an underlying rating of BBB, which effectively had no market, to a rating of AAA, which allowed Flagstar to sell the securitizations and pay lower interest rates to the bondholders. Trial Tr. 39:9–40:10; Roslin Dep. Tr. 196:4–23.

The parties' agreements relating to the Trusts were memorialized in a set of three, simultaneously-executed contracts: the Sale and Servicing Agreements (“SSAs”), see Pl. Ex. 198; Def. Ex. AAI, the Mortgage Loan Purchasing Agreements (“MLPAs”), see Pl. Ex. 451; Def. Ex. AAA, and the Insurance and Indemnity Agreements (“I & Is”), see Pl. Ex. 90; Def. Ex. AAO, (collectively, the “Transaction Documents”). The relevant terms of the 2005 and 2006 versions of these documents were materially identical to one another and will be so treated herein. See Trial Tr. 51:16, 56:25, 64:7–9.

Flagstar served as the sponsor, servicer, depositor, and originator of the loans underlying both Trusts. JPPO ¶ A. 17. After originating the loans, Flagstar Bank sold the loan pools to Flagstar ABS LLC, a subsidiary, through the MLPAs. Trial Tr. 47:24–48:4; Pl. Ex. 451. The SSAs describe the sale of the assets to the securitization Trusts and establish Flagstar's ongoing servicing responsibilities. Trial Tr. 52:1–3; Pl. Ex. 198. The I & Is set out the terms upon which Assured would provide insurance for the transactions and establish a direct contractual relationship between Assured and Flagstar Bank. Trial Tr. 60:19–21; Pl. Ex. 90.

Included in the MLPAs are approximately 75 representations and warranties relating to the loans underlying the pools. See Pl. Ex. 451 § 3.02(a). At trial, the parties relied primarily on two such representations as relevant to the dispute at hand: (1) “Each Mortgage Loan was originated in good faith and in accordance with [Flagstar's] underwriting guidelines”; and (2) “No error, omission, misrepresentation, negligence, fraud or similar occurrence with respect to a Mortgage Loan has taken place on the part of any person, including, without limitation, the Mortgagor, any appraiser, any builder or developer, or any other party involved in the origination of the Mortgage Loan or in the application of any insurance in relation to such Mortgage Loan.” Pl. Ex. 451 § 3.02(a)(36), (65). The SSAs incorporated Flagstar's representations and warranties from the MPLAs. See Pl. Ex. 198 § 2.04(a).

Although Assured is not a party to the MLPAs or the SSAs, sections 7.08 of the MLPAs and 8.06 of the SSAs designate Assured, as the note insurer, as a third-party beneficiary of those agreements. See Pl. Exs. 198 § 8.06; 451 § 7.08. The I & Is also expressly incorporate the representations and warranties made by Flagstar in the SSAs and MLPAs. See Pl. Ex. 90 § 2.03(h). Flagstar further represented in the I & Is that “[e]ach of the representations and warranties of [Flagstar] ... contained in the Transaction Documents is true and correct in all material respects and [Flagstar] ... hereby makes each such representation and warranty to, and for the benefit of, [Assured] as if the same were set forth in full herein.” Pl. Ex. 90 § 2.02(h). Flagstar further agreed in the I & Is that as a “Condition Precedent to the Issuance of the Policy,” Flagstar's “representations and warranties ... in this Agreement shall be true and correct in all material respects as of the Date of Issuance.” Id. at App. A at 42.

The SSAs make Flagstar liable for any material breach of the representations and warranties contained in the MLPAs, regardless of whether or not Flagstar knew that “[its] substance ... was inaccurate at the time [it] was made.” See Pl. Ex. 198 § 2.04(b). The SSAs further require that if any party to the transactions “discovers a breach of any of the foregoing representations and warranties ... that materially and adversely affects the interests of the Trust, the Indenture Trustee under the Indenture, the Noteholders, or the Note Insurer in the Mortgage Loan, the party discovering the breach shall give prompt notice to the other parties and the Note Insurer.” Pl. Ex. 198 § 2.04(c).

The SSAs also require that Flagstar “use all reasonable efforts to cure in all material respects any breach ... within 90 days of becoming aware of it.” Id. § 2.04(d). Should Flagstar breach the representations and warranties and fail to cure such a breach, the MLPAs provide that:

The sole remedy of ... [Assured] ... against Flagstar for the breach of a representation or warranty with respect to a Mortgage Loan ... is [Flagstar' s] obligation, subject to certain cure periods, to accept a transfer of a Mortgage Loan as to which a breach has occurred and is continuing ... or to substitute an Eligible Substitute Mortgage Loan [for the defective loan].
Pl. Ex. 451 § 3.02(c). Like the MLPAs, the SSAs specify that “[Assured's] sole remedy against [Flagstar] for the breach of a representation or warranty with respect to a Mortgage Loan ... is [Flagstar's] obligation ... to accept a transfer of a Mortgage Loan as to which a breach has occurred ... or to substitute an Eligible Substitute Mortgage Loan.” Pl. Ex. 198 § 2.04(e).

Once each Trust was created, bondholders then purchased shares of the Trust and were entitled to payments of principal and interest from the mortgage payments that flowed into the Trust. The Trusts were “over-collateralized,” meaning that there were more loans in the Trusts (and thus greater payments were intended to flow into the Trusts) than were required to meet the payments of the bondholders, so that the Trusts had some excess funding in the event that some borrowers defaulted on their loans. Trial Tr. 807:9–12. Should this cushion fail, Assured guaranteed payments to the bondholders of both the 2005–1 and 2006–2 Trusts in exchange for receiving a premium from such cash flows as might be generated by the underlying loans. JPPO ¶ A. 16.

Flagstar originated the loans underlying these transactions through retail, broker, and correspondent channels. Trial Tr. 871:21–25. Loans originated through the “retail” channel were originated by a Flagstar loan officer. Id. at 872:17–18. Where a real estate broker originated the loan, Flagstar funded the loan prior to closing. Id. at 872:18–20. Loans originated through a “correspondent” were originated by entities, such as credit unions, using their own funds, but even then were subject to Flagstar oversight before being included in the Trust pools. Id. at 872:21–24. As Marni Scott, Flagstar's Executive Vice President of Mortgage Credit Operations, testified, even when loans were not originated by Flagstar loan officers, Flagstar conducted due diligence on the brokers and correspondents through which it obtained loans, including site visits and reviews of the entities' licensing credentials and financial statements. Id. at 874:1–15.

Regardless of the source of the loans, Flagstar underwrote all of the loans in the Trusts to its own guidelines. Trial Tr. 873:4–21. Mortgage underwriting is the process of evaluating against guidelines established by a lender the collateral underlying a loan as well as a borrower's income, assets, and credit. Id. at 300:4–9. Mortgage underwriting guidelines control risk by requiring the collection of documentation to support a borrower's ability to repay the loan and setting limitations on various factors relevant to a borrower's risk of default. Id. at 300:11, 306:18–25.

According to Scott, Flagstar employed approximately fifty underwriters who underwrote HELOCs at the relevant time. Trial Tr. 877:1–3. Flagstar's underwriters had an average of eight or nine years' experience, with a minimum of three years' experience. Id. at 878:7–13. Flagstar provided its underwriters with training on, inter alia, appraisals, fraud detection, and qualifying income, and provided a number of tools for fraud detection and income verification. Id. at 906:4–908:24; 911:10–912:6.

Within the loan pools for the 2005–1 and 2006–2 Trusts were four different types of loans. The 2006–2 Trust consisted of: 43%, full-documentation loans; 0.2%, alternative-documentation loans, in which the borrower provided supporting documentation, rather than the underwriter seeking verification from a third party; 49%, stated-income/verified-asset loans, in which only the borrower's assets were subject to full verification; and 8%, stated-income/stated-asset loans (or “other limited documentation” loans), in which only limited verification was made of the borrower's claimed income and assets. See Trial Tr. 1241:16–1244:12; Def. Ex. AAJ, Annex I–9. The 2005–1 Trust consisted of: 65%, full- or alternative-documentation loans; 34%, stated-income/verified-asset loans, and 0.7%, stated-income/stated-asset loans. See Trial Tr. 1245:1–17; Def. Ex. AAK, Annex I–8. Flagstar's underwriting guidelines, Pl. Ex. 96, applied differently to these four types of loans, and also were different in different time periods. Trial Tr. 519:13–520:20. However, many of the basic principles of Flagstar's underwriting guidelines were constant.

Flagstar's guidelines measured a borrower's capacity to repay a loan by looking at the borrower's debt-to-income ratio (“DTI”), which it calculated by adding a borrower's new mortgage payment to the borrower's already existing recurring debt payments and then dividing that total amount by the borrower's income. Trial Tr. 307:21–24. Flagstar's guidelines set the upper limit for approvable DTI at around 40–50% of the borrower's pretax monthly income. Id. at 308:2–4. In the case of stated-income loans, Flagstar's guidelines further required the underwriter to determine whether the borrower's income is reasonable. Id. at 321:17–24. Thus, Flagstar's product description for HELOC loans provided:

Income must be reasonable for the [borrower's] profession as determined solely by the underwriter. Underwriter will use outside sources to determine reasonable income. One of these sources may be but will not be limited to salary.com.
Pl. Ex. 96, Doc. No. 5557 at 7. Salary.com is an online salary engine that provides salary figures for individuals in a given position in a particular geographic region. See Trial Tr. 338:12–21. Flagstar's guidelines also require that, for stated-income loans, the underwriter need not verify the borrower's claimed income against pay stubs or tax forms, as might be required for a full-documentation loan. Id. at 881:22–882:1. Instead, the underwriter must call each borrower's employer to obtain a verbal verification of employment (“VVOE”) from the borrower's employer, confirming not the borrower's income, but that he is in fact employed at that location and in the position stated in the application. Id. at 322:11–19, 882:19–25.

Scott testified that Flagstar did not always require that underwriters document in the loan file all the verifications conducted in reviewing a loan application, includingVVOE forms. Trial Tr. 918:19–919:5. Nor were underwriters required to reverify a borrower's income where the income was clearly reasonable for the stated employment. Id. at 935:9–17. However, Scott acknowledged that any documentation that the guidelines required an underwriter to obtain should have been kept in the loan file. Id. at 939:5–11.

Flagstar's guidelines evaluated a borrower's credit and payment history by looking at the borrower's Fair Isaac Corporation (“FICO”) score, which provides a numerical summary of the credit history information on a borrower's credit report. Trial Tr. 308:13–17. Beyond the borrower's credit history, a credit report also provides information about the borrower's address, employment history, and any recent credit inquiries conducted by third parties. Id. at 353:7–13. Flagstar's guidelines required a borrower to have a credit score in the range of 620–700 to qualify for a HELOC loan. Id. at 308:19. For borrowers with scores in the range of 620–660, Scott testified that Flagstar's guidelines required that the combined loan-to-value ratio on the mortgaged property, discussed below, be lower than would be required of borrowers in the 660–700 range. Id. at 869:15–21.

Flagstar's guidelines provided for the evaluation of the value of the mortgaged collateral in two steps:

First, the guidelines required that the value of the property be determined through an appraisal, except where state law allowed certain documentation to be used instead to determine the value of the property. Trial Tr. 309:5–9; 382:17–24. In both Flagstar's original underwriting of the loans and the re-underwriting undertaken in connection with this litigation (see below), the underwriters used Automated Valuation Models (“AVMs”): programs that provide real estate property valuations using a mathematical model and data on property values in a given geographic area. Flagstar also employed field appraisal reviews from outside appraisers, which were reviewed by Flagstar's staff appraisers and were verified against an AVM. Id. at 909:7–910:7.

Second, Flagstar's guidelines determined the riskiness of the loan by looking at, with respect to HELOCs, the combined loan-to-value ratio (“CLTV”), calculated by adding the amounts of the first and second lien and comparing that total to the value of the collateral. Trial Tr. 309:1–4. For a loan to be eligible to be securitized by the government-sponsored entities that guarantee mortgages, notably Fannie Mae and Freddie Mac, loan-to value ratio was generally required to be no more than 80%. Roslin Dep. Tr. 26:3–8. However, Flagstar's guidelines permitted a higher CLTV for the HELOCs that were included in the 2005–1 and 2006–2 Trusts. See id. at 27:2–28:10.

Flagstar's guidelines also required that borrowers have cash reserves of sufficient amount so that, in the case of unforeseen circumstances, the borrowers would be able to cover their monthly mortgage payments for a certain number of months after closing. Trial Tr. 309:15–16, 324:24–325:4.

Finally, Flagstar's guidelines allowed for the consideration of compensating factors: that is, where there was evidence in a loan file that the borrower had certain particularly creditworthy characteristics, such as substantial equity in the property (and thus a low CLTV ratio), a particularly high FICO score, or extensive liquid assets, a failure to meet Flagstar's guidelines on another factor might be forgiven. Trial Tr. 916:10–917:2. Flagstar's line underwriters had discretion to consider compensating factors to a certain extent, but substantial exceptions required the approval of a manager or senior leader. Id. at 917:3–21.

At trial, Assured introduced the deposition testimony of Jean Garrick, the head of quality control at Flagstar, who testified that in 2005–2007 Flagstar hired Wetzel Trott, Inc., a third-party auditor, to review samples of Flagstar's approved mortgage loan files for, inter alia, compliance with Flagstar's underwriting guidelines. Garrick Dep. Tr. 9:21–25, 20:3–5. Although it is unclear how many of the loans that Wetzel Trott reviewed were HELOC loans, memoranda from Garrick and others indicate that Wetzel Trott concluded that a fair number of the loan files reviewed had “significant findings” (a euphemism for significant problems). Id. at 59:11–20; see also Pl. Ex. 132 (concluding that “[o]f the 1,209 Conforming, HELOCs–Seconds and SISA loans” reviewed from the third quarter of 2006, “there were 75 loans that had a significant finding for a total of 6.2%,” including 34 loans with underwriting-related significant findings); Pl. Ex. 133 (concluding that 25 of “the 762 Conforming, HELOCs–Seconds and SISA loans” reviewed from the fourth quarter of 2006 contained significant findings related to underwriting).

Stanley Jursek, Flagstar Bank's Executive Vice President and Treasurer and the head of Flagstar Capital Markets Corp., testified at trial that Flagstar's core business is “agency” mortgage loan origination, i.e., underwriting mortgages that conform to the guidelines set forth by Fannie Mae and Freddie Mac. Trial Tr. 800:11–12, 803:2–15. The 2005–1 Trust was Flagstar's first non-agency securitization, and so Flagstar brought in JPMorgan Securities to advise them on the transaction. Id. at 805:24–806:9. Jursek also testified that Flagstar deemed loans inappropriate for securitization if they had a history of delinquency or were not underwritten using Flagstar's automated underwriting system. Id. at 821:23–822:5.

Assured, for its part, analyzed each of the Trusts before deciding to insure the securitizations to determine what risks Assured might face. Trial Tr. 70:23–25. Assured evaluated Flagstar's lending and servicing operations and engaged in a due diligence review of the loans underlying each transaction. Assured's operational review included on-site visits and meetings with Flagstar senior management responsible for loan origination and servicing, in order to assess the quality of Flagstar's controls and thus how confident Assured should be in its loss predictions. Id. at 125:17–25, 127:1–4, 128:1–5.

Assured's due diligence process consisted of both a file review of a sample of the loans underlying each pool and a review of each pool's loan data tape, which lists key data points about each loan to be included in the pool ( e.g., borrower name, address, income, assets and credit score, collateral value, LTV, and DTI ratios). Trial Tr. 823:13–824:3. George Stiehl, former-Vice President in FSA's Residential Mortgage Group (the unit responsible for RMBS at the time), testified at his deposition that Assured engaged in the due diligence process in order to be able to predict and plan for expected losses on the two transactions. Stiehl Dep. Tr. 12:10, 15:11, 109:10–13. However, even after conducting its diligence review, Assured insisted on obtaining, and did obtain, the contractual representations and warranties from Flagstar. As Russell Brewer, Assured's Chief Surveillance Officer and one of the members of the Management Review Committee that approved Assured's participation in the Flagstar transactions, testified, Assured would not have insured the securitizations had it not received the representations and warranties in the Transaction Documents. Trial Tr. 63:19–24. This was material to Assured because, interalia, Assured would have been unable to appropriately price the transaction's risk level if it could not rely on the quality of the underwriting of the underlying loans. Id. at 45:2–10, 49:21–25, 51:4–7.

Put another way, Assured's due diligence review was limited in important respects. It consisted of selecting samples of the loans to be included in the transactions, which were then re-underwritten to Flagstar's underwriting guidelines and graded to Assured's internal guidelines. Stiehl Dep. Tr. 64:13, 68:6, 172:16–22. The Clayton Group (“Clayton”), a third-party due diligence provider, performed the loan file due diligence review on the adverse and random samples selected from the 2005–1 loan pool, and the Bohan Group (“Bohan”), another third-party provider, performed the loan file review on the random sample selected from the 2006–2 pool. Trial Tr. 71:1–7, 77:23–25. The due diligence reviews did not look for evidence of fraud, nor did Clayton and Bohan utilize any third-party fraud detection tools, and Clayton and Bohan were only expected to flag issues if something in the loan files appeared “ridiculous.” Trial Tr. 224:10–13, 1108:17–21; Stiehl Dep. Tr. 77:23–24; see also Def. Ex. BAN ¶ 59 & tbl. 4 (citing no fraud-related findings).

In comparing the loan files against Flagstar's guidelines, Clayton and Bohan graded the loans according to a coding system: loans were classified as Event Level 1, Event Level 2, or Event Level 3, with Event Level 3 being the most problematic. See Pl. Ex. 99 at 3 (defining Event Levels). The Event Level 1 classification covered loans that met the stated underwriting criteria and were in compliance with all federal and local lending laws. Id. Loans classified as Event Level 2 presented an exception to the underwriting guidelines, but the reviewer identified adequate compensating factors to justify the exception, making it immaterial. Id. Loans coded as Event Level 3, effectively those that were deemed to be materially noncompliant, see Trial Tr. 1037:1–8, presented one of four issues: the loan failed to meet Flagstar's guidelines, and acceptable compensating factors were not present; the property was unacceptable; the loan was not made in compliance with federal and local lending requirements; or the facts and circumstances of the loan would lead the underwriter to believe the borrower's situation would not improve with the loan. Pl. Ex. 99 at 3.

Over the course of their reviews, Clayton and Bohan engaged in iterative processes with Flagstar and Assured, in which they were able to seek additional documentation and address compensating factors within the loan files. See Stiehl Dep. Tr. 96:4–16. This “clearing” process affected the grades given to the loans in the samples, as the due diligence providers were able to clarify issues with Flagstar and obtain documentation that might not have been located in the HELOC loan file ( e.g., documentation was located in the file for a first-lien mortgage when Flagstar granted the borrower a first-lien and a HELOC loan contemporaneously). See Trial Tr. 1035:15–1036:8. By this process, nearly all of the loans originally coded as Event Level 3 were cleared of material issues. See id. at 1032:7–1033:2; 1035:19–1036:8.

After the due diligence reviews concluded, Assured's Residential Mortgage Group prepared executive summaries for each transaction and its Management Review Committee approved the transactions based on those summaries. Trial Tr. 74:20–75:17; Pl. Exs. 99, 100. The executive summary for the 2005–1 Trust reported that “[o]f the 125 loans that were reviewed, 0 loans were coded as Event Level 3 for [a] reason other than the TILA violation [a regulatory violation of no relevance to this case]. Clayton is in the process of clearing credit and compliance exceptions.” Pl. Ex. 99 at 3. A related memorandum on Clayton's loan file due diligence reported that “[t]he file review of the Original Random Sample show these loans to be excellent credit.” Id. (Mem. dated Nov. 21, 2005 at 1). Similarly, the executive summary for the 2006–2 Trust reported that “[o]f the 250 loans that were reviewed, 6 were coded as Event Level 3 due to trailing document and [right of rescission] issues. These loans are in the process of being cleared. It is Flagstar's belief that all of the loans will be cleared prior to the closing of the transaction.” Pl. Ex. 100 at 3.

In grading the loans against Assured's internal guidelines, Clayton and Bohan scored the sample of loans on credit, ability to pay, and collateral quality. Trial Tr. 129:1–12. The credit grade was based on reported delinquencies in a borrower's credit report; ability to pay, on the borrower's DTI ratio and liquid assets; and collateral, on the property's LTV ratio and the kind of occupancy of the property. Stiehl Dep. Tr. 72:24–74:6. The Residential Mortgage Group's executive summaries for the two securitized pools reported grades for both random samples of “A+” for credit quality, “A-” for ability to pay, and “C” for collateral. Pl. Exs. 99 at 3; 100 at 2; Trial Tr. 156:16–23, 181:18–182:10. Brewer testified at trial that the collateral underlying the loans was given the grade of “C” because HELOCs tend to have high CLTV ratios as second-lien products. Trial Tr. 159:9–15.

In the loan tape reviews, Clayton and Bohan verified the integrity of the tape by checking the data reported on the loan tape against the information in the loan files included in the due diligence re-underwriting sample. Trial Tr. 123:1–124:6.

Assured then used the data from the loan tape and the results of the due diligence review to model its expected losses on the transactions and determine thereby whether the transactions would be appropriate for Assured to insure. Stiehl Dep. Tr. 12:17–13:13. In this case, Assured entered the data into a few different models that looked at different loan characteristics and decided to average the outputs of those models, deriving an average expected loss rate of around four percent.⁴ See id. at 12:17–13:13, 124:6–10; Beard Dep. Tr. 238:13–20; Williams Dep. Tr. 226:25.

4 Although Assured's modeling is useful as background, the issue—debated at trial, see, e.g., Trial Tr. 82:12–25—of whether Assured failed to model the transactions correctly and therefore ended up paying more than it planned does not speak to whether Flagstar's alleged misrepresentations caused an increased risk of loss on the loans, as Assured's models were based on the assumption that Flagstar had complied with the representations and warranties in the Transaction Documents.

At the time of approving the transactions, Assured also understood that there were a variety of additional risk factors that could affect the performance of the securitizations. For example, Assured's Management Review Committee discussed the risk of a real estate “bubble,” which, if it burst, could mean that housing prices might drop in the future. Trial Tr. 116:18; Def. Ex. AYQ. Second, Assured discussed risks associated with the fact that the underlying loans were ten-year “bullet loans,” i.e., loans that were interest-only for ten years, at which point the entirety of the principal would become due, a risk that would occur at the same time for many of the loans in these transactions. Trial Tr. 44:2–9; see also Pl. Exs. 99 at 2; 100 at 1. Third, Assured recognized that HELOCs, as second-lien products, are subordinated to first-lien mortgages; should a homeowner default as property values fell, a holder of a second lien would likely be unable to recover on any of the collateral underlying the mortgage, as the first lien would have to be satisfied in its entirety before a second lienor could receive any value from the collateral. Trial Tr. 159:9–15. Finally, Brewer testified that Assured generally looks at the geographic concentration of borrowers as a risk factor for a securitization, since regional economic downturns may have a disproportionate effect on the performance of the security. Trial Tr. 163:12–164:14. In the 2005–1 Trust, over fifty percent of the loans were on properties located in California, Michigan, or Florida, although it is contested whether this qualifies as geographically concentrated. Id. at 163:20–21, 199:25.

Assured has paid about $14.7 million in claims arising from the 2005–1 Trust and approximately $75.4 million in claims arising from the 2006–2 Trust. Pl. Exs. 458 at 4, 459 at 5. At this point, Flagstar has reimbursed Assured only for about $950,000 of the claims Assured paid to bondholders, see Trial Tr. 446:20–447:24, and Assured may be responsible for additional claim payments in the future. Id. at 43:19–25.⁵

5 Flagstar's argument that, as to the 2005–1 Trust, Assured will eventually be reimbursed by Flagstar pursuant to provisions that subordinate Flagstar's ultimate interest in the loans to Assured's interest is discussed below.

By two letters dated January 13, 2009 and two letters dated July 9, 2010, Assured made formal repurchase demands on both the 2005–1 and 2006–2 transactions. JPPO ¶¶ A. 34–35; Pl. Exs. 464, 465. To date, however, Flagstar has repurchased only five loans from the 2005–1 Trust, not in response to a particular demand, and two loans from the 2006–2 Trust after discovering evidence of fraud in those loans. Trial Tr. 928:12–929:12, 930:2; see also Pl. Ex. 407 (listing Flagstar's repurchases).

THE WAR OF EXPERTS

Given the aforementioned facts, this case essentially reduces to resolution of conflicting expert testimony. The expert testimony is summarized as follows:

Dr. Nelson Lipshutz, Assured's expert on statistical sampling,⁶ created a random sample of 400 loans from each of the two securitization pools, for a total of 800 loans. Trial Tr. 253:8–15. In determining the size of the random samples for the two Trusts, Dr. Lipshutz applied a formula that took into account the characteristics of the overall loan pools, the purpose for which the sample would be used, and the objective that “the estimate derived from the sample has a 95% chance of being within ... plus or minus five percentage points of the actual population proportion,” a standard measure for a high level of statistical confidence. Id. at 259:16–260:2. In particular, Dr. Lipshutz designed the samples with the understanding that they would be used “to make a binary decision on each loan,” that is, whether the loan file conforms to the representations and warranties made in the Transaction Documents, or not. Id. at 257:8–12.

6 Dr. Lipshutz is the President of Regulatory Research Corp., a statistical consulting firm. Trial Tr. 245:22–24. He has a doctoral degree in physics and a subsequent master's degree in business administration. Id. at 246:12–14.

Here, the chosen sample size of 400 loans per loan pool was slightly larger than the minimum number of loans that Dr. Lipshutz determined would be necessary to obtain a representative sample for each pool—371 loans for the 2005–1 pool, and 358 for the 2006–2 pool. Trial Tr. 254:3–19. Dr. Lipshutz then confirmed the representativeness of the samples by testing the distribution in the sample of seven relevant variables ⁷ (provided by Rebecca Walzak, Assured's mortgage underwriting and origination expert) against their distribution in the overall populations to ensure that there were no statistically significant differences in distribution. Id. at 260:11–21, 275:21. He did not, however, test the representativeness of the sample as to loan payment status ( i.e., whether the borrower was delinquent on payments) and original principal balance of the loans. See id. at 261:2–8. Dr. Lipshutz concluded that the random samples are representative of the total loan populations in both the 2005–1 and 2006–2 loan pools. See id. at 261:22–262:6.

7 Those variables were distribution by state, loan-to-value ratio, debt-to-income ratio, outstanding principal balance, note rate, appraisal type, and credit score. Trial Tr. 261:2–8.

Assured's mortgage origination and underwriting expert, Rebecca Walzak, ⁸ reviewed the loan files for the 800 loans in Dr. Lipshutz's random sample in order “to determine if the loans in these securities complied with the representations and warranties” in the Transaction Documents. Trial Tr. 304:9–22, 305:7–306:4. Walzak testified that she was instructed by plaintiff's counsel that the standard for considering a loan to be “defective” was whether, at the time the loan closed, there was a breach of the contractual representations and warranties—specifically, whether there was either a failure to follow Flagstar's underwriting guidelines or fraud or misrepresentation in connection with the loan—that materially increased the risk of loss to Assured. Id. at 305:13–15. Walzak concluded that 606 of the 800 loans in the sample contained material breaches of Flagstar's representations and warranties. Although Walzak originally found four additional loans from the 2006–2 transaction to be materially defective, in reviewing the loans for trial she determined that, while these loans breached the representations and warranties in the Transaction Documents, those breaches were not severe enough to deem the loans materially defective. See id. at 315:7–10. As broken down between the two transactions, Walzak found that 340 loans in the 2005–1 sample (85%) were defective, and 266 (originally 270) loans in the 2006–2 sample (66.5%) were defective. Id. at 315:2–4. Walzak also found evidence of fraud in 14% of the loans in the 2005–1 sample and 19.75% of the loans in the 2006–2 sample. Id. at 316:22–24, 533:5–11.

8 Walzak is the President of RJB Walzak Consulting and has worked in the mortgage industry since 1979. Trial Tr. 298:23, 299:17. Walzak began her career as a mortgage underwriter at Ryan Financial Services, and she has since focused on quality control and risk management relating to mortgage underwriting. See id. at 299:16–302:16. She holds a master's degree in business administration and a certificate in quality management. Id. at 299:12–15.

Walzak created summary spreadsheets for each of the 800 loans in the sample, which presented her core findings on each loan. See Pl. Exs. 189 (2005–1 analysis); 190 (2006–2 analysis). The spreadsheets include 106 yes/no questions in nine different areas of underwriting,⁹ some of which addressed compliance with Flagstar's guidelines, while others addressed the execution of the loan underwriting process itself. Trial Tr. 315:19–22, 392:22–393:2. While some of the questions could be answered on a yes/no basis—e.g., whether required documentation was included in the file—others required a determination of reasonableness—e.g., whether a borrower's stated income was reasonable for his position. Trial Tr. 395:1–9. The final three questions were designed to be summary questions. Question U–124 read: “Was all the documentation and/or explanationsrequired by the underwriter obtained?” See Pl. Ex. 189. Question U–125 read: “Did the loan comply with the appropriate underwriting guidelines?” Id. The final question, U–126, was awkwardly worded. It asked: “Did the identified defects in the loan file not materially and adversely affect Assured's interests in the loan?” Id.; Trial Tr. 498:16–19. Walzak testified that the question was worded for counting convenience, so that here, as elsewhere, a “No” answer corresponded to an underwriting problem. Trial Tr. 498:20–23, 553:17–18. Walzak further testified that, in any case, she was the one who ultimately made the materiality decision based on all the findings she received. Id. at 554:5, 557:24–558:1.

9 Those areas were: assets, borrower credit, employment, income, fraud, property, title/closing, regulatory issues, and overall underwriting practices. See Trial Tr. 1020:17–20.

As to Walzak's process, she was assisted in her review of the loan samples by two separate groups: a team of eight experienced underwriters and a supervising underwriter, who conducted the initial review of the 800 loan files, and third-party consultants, Digital Risk, LLC, and Clear Capital, Inc., which looked for fraud and misrepresentations in the loan files and appraisals. Trial Tr. 306:7–11, 379:11. As for the underwriting team, Walzak instructed the underwriters to review the loan files (provided by Flagstar) to determine if they complied with Flagstar's underwriting guidelines, which Walzak provided to them. Id. at 306:16–17; 518:14–15. The reviewers then re-underwrote each loan to the guidelines, filled out the 106–question spreadsheet, and entered comments on their most significant findings before sending the results of their review and the loan file back to Walzak. Id. at 310:10–12, 488:9; see also Pl. Exs. 189, 190. The reviewers returned their findings and the loan files to Walzak on a rolling basis as they completed batches of loans. Id. at 488:12–20.

The underwriting review protocol Walzak provided to her reviewers described the standards to be applied in a half-page paragraph. See Def. Ex. AZP. At the outset, Walzak conducted a one-hour conference call, in which she and her underwriting team discussed the instructions—which covered multiple sets of underwriting guidelines and the 106 individual questions on the spreadsheet—and she answered any initial questions from the reviewers. Trial Tr. 523:1–8. The underwriting team was spread throughout the country, and Walzak primarily communicated with the reviewers only indirectly, through emails with the supervising underwriter, although she held weekly conference calls with the underwriters. Id. at 516:18, 517:23, 591:4–23.

As discussed above, Flagstar's underwriting guidelines allowed some discretion to the individual underwriter to determine whether certain borrower characteristics fell within the realm of “reasonableness.” In order to offer additional guidance to her underwriting team, Walzak provided them the results of a survey she had conducted in 2007, which polled twenty experienced underwriters to determine whether there were common industry standards as to what is considered reasonable with respect to a variety of underwriting factors. Trial Tr. 407:12–14, 505:3.

The loan file review occurred over a four-week period in October and November 2011. Trial Tr. 583:19–584:10. Walzak testified that early in the process, she reviewed ten to fifteen percent of the loan files in their entirety to ensure that the underwriting team had a consistent understanding of what the guidelines required, so that she could address any disparities among reviewers' understanding of the questions. Id. at 489:2–9. When, as the reviewers' analyses came back, an underwriter indicated that there were no material issues in the file, Walzak put that file aside and conducted no further review of the loan. Id. at 310:21–25. For the loans in which a problem was indicated, Walzak made the ultimate determination of whether a loan was materially defective. Id. at 310:21–25. Although Walzak's initial testimony as to the extent of this review was somewhat unclear, see id. at 493:4–7, the Court credits her ultimate testimony that she in fact reviewed each loan file in its entirety, see id. at 401:9, albeit within the confines of the four-week window in which her review needed to be conducted.

As for the fraud review, Walzak relied on Digital Risk, a third-party vendor, to look for evidence of borrower fraud or misrepresentations, especially as to borrower income, debts, and occupancy status. Trial Tr. 306:10–11. Digital Risk reviewed the loans for fraud using public records, including DMV records and bankruptcy filings, online salary engines likesalary.com and payscale.com, the Work Number (an employment verification database), and information from the U.S. Department of Labor's Bureau of Labor Statistics. Id. at 306:10–17. When Digital Risk discovered information that appeared inconsistent with that reported in the loan files, Walzak then examined the information personally to determine whether there was a significant enough discrepancy to find the loan defective. Id. at 311:24–312:2.

Digital Risk in turn subcontracted to another third-party vendor, Clear Capital, to review the appropriateness of the appraisal values of the subject properties. Trial Tr. 379:11. Clear Capital used its proprietary AVM (Automated Valuation Model) to compare the subject property to similar properties in the same neighborhood (“comps”) that sold within a year of the subject property, in order to determine whether the value Flagstar assigned to the subject property was reasonable. Id. at 327:1–7, 381:8–11. Clear Capital worked solely through Digital Risk, providing their new appraisal results to Digital Risk directly, and Walzak never spoke directly with anyone at Clear Capital about their work on this assignment. Id. at 379:11–380:9.

Walzak's overall findings were as follows: first, 86 borrowers had undisclosed debts that were not reflected in the loan file or were not included in the Flagstar underwriter's calculation of DTI, and which, when factored in, contributed to a DTI above the limit allowed by Flagstar's guidelines. Trial Tr. 321:6–15. Walzak included in her findings undisclosed loans that closed within 30–60 days after the closing of the subject loan, which would have been discoverable by the original Flagstar underwriter through inquiries on the borrower's credit report. Id. at 677:5–7. Walzak claimed that the Flagstar guidelines required a letter of explanation from the borrower to address such inquiries, id. at 314:10–16, although Marni Scott, Flagstar's head of underwriting, testified that Flagstar's guidelines did not require letters of explanation for HELOCs at the time that the loans in the securitizations were underwritten, nor did Flagstar require its underwriters to check for debts obtained after closing. Id. at 915:6, 916:4. Walzak separately counted 236 loan files in which there were unexplained recent credit inquiries on the borrower's credit report. Id. at 323:22–324:11.

Walzak found 131 instances of unreasonable borrower income as claimed in stated-income loans. Trial Tr. 322:4–10. To make this determination, Walzak instructed her reviewers to use an online salary engine, such as salary.com, salaryexpert.com, or payscale.com, to determine the range of income a borrower could be expected to make in his stated position and geographic area. Id. at 338:14–21. In order to give the benefit of the doubt to Flagstar, Walzak instructed her reviewers to assume that the borrower was in the top ten percent of earners in his position, and thus had them compare the borrower's stated income against the ninetieth percentile reported for that area. Id. at 350:9–11. If the stated income exceeded the ninetieth percentile, Walzak treated the stated income as unreasonable. Walzak also had her reviewers use data provided by the U.S. Department of Labor's Bureau of Labor Statistics when possible. Id. at 567:6–11. Walzak and later witnesses explained that, although online salary engines include information on a wider range of positions, they do not use historical data and thus risk understating 2004–2006 incomes given the economic decline of the intervening years, while the Bureau of Labor Statistics includes such historical information. Id. at 567:6–11. Walzak admitted that salary engines were used in the majority of cases, id. at 603:22, but testified that potentially higher salaries in 2004–2006 were not an issue because the discrepancies between the borrowers' reported incomes and the reported ninetieth percentile were so large that it was apparent that the borrowers had inflated their income. Id. at 567:9–22. However, where Walzak was able to obtain information from the borrower's employer or other documentation regarding the borrower's actual salary, she made a finding of borrower fraud (an automatic breach of the representations and warranties), which occurred in 10 instances. Id. at 606:1–607:1. Walzak also counted 208 instances in which the loan files were missing a VVOE form, and assumed, in the absence of any evidence to the contrary, that no such verification of employment occurred. Id. at 323:6–7.

Where there was a material variance between a borrower's income as stated in his loan application and the ninetieth percentile as reported by the salary engine, Walzak instructed the loan reviewers to recalculate the borrower's DTI ratio. Id. at 604:16–18. Undisclosed debts, unreasonable stated income and revised DTI ratios exceeding Flagstar's guidelines were treated as separate violations of Flagstar's underwriting guidelines. Id. at 606:1–3. Walzak found 125 loans in which DTI was miscalculated (and in which her revised calculation exceeded Flagstar's guidelines) based on a misstated income, undisclosed debt, or simply an error in the original calculation. Id. at 325:16–326:10.

Based on the information provided by Clear Capital, Walzak determined that Flagstar's appraisal underlying the stated value of the property was flawed for 24 loans. Trial Tr. 326:23–327:17. Upon revaluing the property, Walzak also recalculated the CLTV ratio. In 18 instances, the adjusted CLTV exceeded the guidelines limit for the relevant loan. Id. at 326:13–18.

Walzak also determined that there were 17 instances in which the borrower misrepresented his occupancy of the property. Trial Tr. 328:15–16. A borrower who is taking a loan on her primary residence is regarded as presenting the least risk, as borrowers tend to protect their homes from foreclosure even under stress. Id. at 328:4–8. A second home, generally a vacation property, is riskier than a primary home from the lender's standpoint, but it is less risky than an investment property, in which the borrower has no stake other than his investment. Id. at 327:20–328:16. Walzak found this to be a significant issue where the borrower's actual use for the property (usually, as an investment property) indicated that the loan was riskier than Flagstar initially determined based on the borrower's misrepresentation (usually, as a primary residence). See id. at 328:11–14.

Finally, Walzak counted 49 instances in which the borrowers had insufficient cash reserves to meet Flagstar's requirements. Trial Tr. 324:22–325:12.

Walzak opined more broadly that Flagstar's underwriting process was uncontrolled and inconsistent, and the loans seemed “as if they were rubber-stamped just to get them through the process.” Trial Tr. 316:13–18. Walzak found that the Flagstar underwriters failed to address “red flags” and indicia of fraud or inconsistencies in 178 loan files, id. at 329:20–330:8; miscalculated the borrower's income in 104 instances, id. at 331:12; failed to include all debts in the DTI calculation 70 times, id. at 331:14–20; and failed to include in the file required documentation, including asset documentation and signed loan application forms, 179 times, id. 331:22–332:4. Although the process-related issues that Walzak identified did not necessarily indicate that a loan was defective, Walzak suggested that such issues spoke to the pervasiveness of the underwriting defects in the loan pools. See id. at 332:7–16.

At trial, Flagstar presented its own underwriting expert, John Griggs, ¹⁰ who testified to his conclusions about the validity of Walzak's loan-level allegations. Trial Tr. 1150:3–5. Griggs did not re-underwrite the loans, but rather reviewed Walzak's determinations for both the validity of the claims and the materiality of the issues identified. Id. at 1150:7–11. Of the 606 loans identified by Walzak as defective, Griggs found that 484 had been repaid in full or were current in their payment status as of October 31, 2011, 2011 WL 5335566.Id. at 1151:25–1152:5. Since Assured seeks damages only as to loans that have defaulted or become delinquent—it is only on these loans that Assured has had to pay its insurance—Griggs therefore focused the rest of his analysis on the remaining 126 loans in Walzak's sample that were either in default or delinquent and that she found were defective in one or more of the ways previously discussed. Id. at 1163:1–2. Griggs opined that Walzak's findings of material breach were supported for only three of the 126 loans, and, more generally, that her findings were without basis or identified only non-material defects. See id. at 1217:12–22.

10 Griggs is the owner and founder of Solutions Associates, a management consulting firm. Trial Tr. 1143:8–25. He holds a master's degree in business administration and has 38 years' experience in consumer financial services. Id. at 1144:9–14. Griggs testified at trial that he has personally underwritten thousands of loans and has conducted thousands of loan-file reviews. Id. at 1147:11–21.

In structuring his review, Griggs hired a team of six underwriters and one manager, who reviewed the loan files from the Washington, D.C. offices of Navigant Consulting, Inc., a consulting firm hired by Flagstar in this case. Trial Tr. 1164:1–4, 1167:8–11. Relying on the loan files, servicing records, Flagstar's underwriting guidelines, and Walzak's work papers, Griggs's team of underwriters evaluated Walzak's 1280 allegations for the 126 loans to determine whether each allegation was accurate. Id. at 1165:15–22. The underwriters then created a summary for each loan with the results of the review on an allegation-by-allegation basis. See id. at 1165:24–1166:2; Def. Ex. BBD 1–126. Griggs himself reviewed the 126 summaries and examined a sample of approximately a dozen loan files in full. Trial Tr. 1265:8.

Griggs scored each allegation within each loan, and then each loan was assigned a category corresponding to the most adverse rating received on any individual allegation for that loan. Trial Tr. 1169:7–1171:2, 1177:18–22. A score of 1 indicated that Walzak failed to establish an issue; in essence, Griggs judged that Walzak got the allegation wrong. A total of thirteen loans of the 126 reviewed received a highest allegation score of 1, four from the 2005–1 sample and nine in the 2006–2 sample. Id. at 1269:13–21. Examples cited by Griggs included: a loan as to which Walzak alleged that the borrower's claimed assets were not supported, but documentation of those assets existed in the loan file, id. at 1186:1–3; a loan as to which Walzak alleged that there was no VVOE form in the file, when no form was required by the guidelines because the borrower lived on social security and pension income, id. at 1186:13–18, 1216:10–13; and a loan as to which Walzak claimed that the income was unreasonable in comparison to the ninetieth percentile on salary.com, when other information in the file, such as credit lines and payment history, indicated that the claimed income might be reasonable, id. at 1187:23–1188:5.

Allegations scored as a 2 were those in which Walzak correctly identified an issue, but that issue was either curable or was not material. Griggs classified a total of 67 loans (28 in the 2005–1 sample and 39 in the 2006–2 sample) as having a score of 2. Trial Tr. 1268:22–1269:2. Griggs's findings of curable or non-material issues related primarily to two issues: First, Griggs treated as curable any allegations that required documentation was missing from the loan file, but for which the paperwork was later provided by Flagstar.¹¹ Id. at 1214:21–22. Second, Griggs treated as not material allegations that the borrower had significantly misrepresented his income if the borrower successfully made payments on the loan for three years. ¹² Id. at 1304:2–15. However, Griggs acknowledged that if he removed this repayment filter, then a finding that a borrower had significantly misrepresented his income would rise to the level of a 3. Id. at 1307:9–16.

11 During discovery, Flagstar provided some of the documentation that was missing from the loan files in a subsequent production. This production occurred after Walzak issued her expert report, but before Griggs issued his. In creating his scores, Griggs treated any claim of missing documentation as “curable,” rather than “cured,” in his expert report, and therefore was bound to treat them as such at trial. See Trial Tr. 1193:3–1200:16.

12 At trial, Flagstar introduced evidence that loans that have performed satisfactorily for a period of time (3 months or a year) are widely assumed by underwriters to be free from origination fraud. See, e.g., Def. Ex. AYX at 38 (Assured making this assumption); Trial Tr. 1152:10–12 (Griggs testifying that any origination errors on loans performing for 6–7 years had no impact on the risk of loss on those loans). This assumption, however, is at best only tangentially relevant as the issue here, as discussed below, is not whether failure to adhere to Flagstar's underwriting guidelines directly caused the loans to default, but rather whether the presence of such errors materially increased the risk of default.

Allegations rated 3 were deemed to be material deficiencies. Trial Tr. 1170:5–7. As mentioned above, Griggs determined that three loans were materially deficient, and he recommended their repurchased to Flagstar. Id. at 1246:2–16. Of these, two loans had undisclosed debts and therefore should not have qualified for a loan. Id. at 1246:18–21. In the third, a husband and wife had agreed to separate their debts upon divorce, in effect agreeing to dis-obligate the other from that debt, which is legally ineffective. Id. at 1247:1–13.

Griggs created three additional categories. Over half of Walzak's allegations were graded “D” for duplicative or derivative findings. Trial Tr. 1178:7–9. Griggs graded an allegation as a D when a “no” answer to one question necessarily resulted in a “no” to a similar question, and he always treated “no” answers to summary questions U–125 and U–126 as duplicative. Id. at 1170:8–413, 1189:4–7. Griggs graded allegations “NC” when he or his reviewersdid not understand a question or did not know why an allegation was being made, and found 30 such allegations to be unclear. Id. at 1169:22–23. He further commented that only 206 comments on Walzak's spreadsheets supported, clarified, or even related to the allegations in the preceding 106 questions. Id. at 1167:1–3, 1192:11–13. Finally, Griggs graded allegations “NR,” or not rated, when he deemed the loans as to which the allegations were made to be “life event” loans.¹³ According to Griggs, “life event” loan are those loans for which, from reviewing the servicing records, he or his reviewers found evidence that a life event unrelated to any underwriting deficiencies significantly changed the borrower's circumstances, such as divorce, unexpected medical expenses, or loss of income. Id. at 1178:17–24, 1179:1–3. Once Griggs concluded that a loan was a life event loan, he did not evaluate the validity of any of Walzak's allegations with respect to that loan. Id. at 1190:24–25. In total, Griggs determined that 43 of the 126 loans that he reviewed were life event loans, 12 in the 2005–1 sample and 31 in the 2006–2 sample. Id. at 1266:20–24.

13 Griggs also rated as “NR” Walzak's other allegations for the three loans in which he found allegations that he scored as 3. See Trial Tr. 1170:17–24.

Griggs also criticized Walzak's decisions more broadly. Griggs found that Walzak's duplicative questions caused her to overstate the number of allegations on each loan, and further opined that Walzak alleged violations of Flagstar's guidelines that were technically correct but were of no consequence to the potential risk of loss for that loan. Trial Tr. 1218:18–22, 1223:19–24. Griggs disapproved of what he saw as Walzak looking at each question in her spreadsheet as an independent event, thereby failing to take compensating factors into account in any of the comments he reviewed, id. at 1158:1–5, 1236:7–9, although, when pressed on cross-examination, Griggs acknowledged that Walzak's questions explicitly referenced compensating factors (specifically, questions U–105 and U–111). Id. at 1340:7–25. Finally, Griggs was critical of Walzak's fraud and misrepresentation claims because they were based on information obtained from online salary engines like payscale.com and salary.com, AVMs, and Accurint, all of which he considered to be inherently unreliable. Id. at 1226:3–25, 1230:12–23, 1235:18–20. As to the salary engines specifically, Griggs testified that the data on which the numbers are based are incomplete, as they tend to contain only that information which companies choose to provide, and they contain no historical information. See id. at 1187:7–19. By contrast, Griggs concluded that he found nothing unusual about the quality of the loans he reviewed, and his review indicated that Flagstar's underwriting process was not uncontrolled, as Walzak testified. Id. at 1259:13–19.

To the extent that Griggs opined that Walzak attempted to find as many allegations as possible without regard to materiality, see Trial Tr. 1219:14–19, 1222:25–1223:6, Griggs appears to have the opposite tendency, as he tended to discount issues in the loan files that Walzak treated as material. In one case in which Walzak found a borrower's stated income to be unreasonable, Griggs discounted an entry in Flagstar's servicing notes that implied that the borrower had lied about being employed at the time he applied for the loan because the borrower had made seventeen payments on the loan before defaulting. See id. at 1317:21–1319:18. Griggs also assumed that where undisclosed debts appeared to be mortgages on investment properties, potential rental income should be taken into consideration in calculating the DTI ratio, even when there was no information in the loan file about any rental agreement or income. See id. at 1221:10–14, 1288:16, 1317:6–7.

Both Walzak and Griggs testified regarding their findings on specific loan files included within the 606 that Walzak found to be defective. Since Griggs reviewed only the 126 defective loans that had already defaulted or were delinquent, his opinions extended only to those loans. In total, the witnesses testified to the details of over twenty loans, and the Court will highlight here about a dozen examples as illustrations.

As to the loan ending in 9016,¹⁴ Walzak found that the borrower had misrepresented both his debt obligations and his income, leading to an incorrect DTI calculation by the Flagstar underwriter. Specifically, Digital Risk discovered through the Mortgage Electronic Registration System (“MERS”) that the borrower, prior to closing on the subject loan, had purchased other properties worth $480,000 and $270,000, on which he had taken out mortgages. As to income, the borrower claimed that he was a private investigator earning $30,000 per month, but the payscale.com report indicated that an income of approximately $15,000 per month (as the ninetieth percentile for the borrower's geographic area) was more reasonable. Walzak's team then recalculated the borrower's DTI as 186.34%, which vastly exceeded Flagstar's guidelines range of around 40–50%. See Trial Tr. 337:5–339:18. Griggs found that the misrepresentation of the borrower's debt obligations was not material because Walzak's recalculated DTI failed to account for potential rental income on the additional properties, as Griggs found it “reasonable to assume the properties in question were investment properties and that the borrowers would have received a rental offset.” Def. Ex. BBD 80 at 69.

14 To protect borrower privacy, the parties identified the loans only by the last four digits of the loan number. The Court adopts the same convention here.

For the loan ending in 6077, Walzak again found that the borrower misrepresented both her debt obligations and her income, leading to an incorrect DTI calculation. The borrower stated that she was a curriculum coordinator at a private school earning $7,335 per month at age 24 and with no college degree, which is usually required for such a position. According to payscale.com, the ninetieth percentile for a curriculum coordinator in the borrower's geographic area—a generous assumption given the borrower's characteristics—was $5,108 per month. As to the debt obligation, the borrower closed on a $110,000 loan on the same property two weeks after closing on the subject loan, and Walzak claimed that Flagstar's underwriter would have seen inquiries on the borrower's credit report at the time of closing that were indicative of this new debt and should have required a letter of explanation. With both of these adjustments, Walzak's team recalculated the DTI to be 108.84%, substantially above Flagstar's guidelines range. See Trial Tr. 339:22–341:20. In response to Walzak's findings, Griggs found that these concerns were not material: the borrower had a high FICO score and additional rental properties; her employment had been verified according to the VVOE forms found in the loan file (and online salary engines are unreliable in any case); and the undisclosed debt closed after the subject loan when the Flagstar guidelines created no obligation to obtain a letter of explanation. See id. at 1255:20–1258:8.

On the loan ending in 2799, Walzak found that the borrower misrepresented both his income and debt obligations. The borrower claimed to be a corporate production manager earning $20,200 per month. Digital Risk not only found that, according to the Bureau of Labor Statistics, the ninetieth percentile income for this position at the time was $11,850 per month, but also verified with the borrower's employer that his income was $3,981 per month, both far shy of the borrower's represented income. Walzak also noted that the borrower had an undisclosed mortgage on another property in the amount of $77,250, as well as seven inquiries on his credit report but no letter of explanation in the borrower's file. See Trial Tr. 348:3–349:10. Griggs did not review this loan.

As to the loan ending in 0641, Walzak found that the borrower had misrepresented his income, debt obligations, and the occupancy of the property. The borrower reported that he was an electrical engineer employed by the local school district for twelve years making $8,850 per month, but Digital Risk reported that the ninetieth percentile income, according to the Bureau of Labor Statistics, was $4,973 for the borrower's area. Moreover, at the time of the closing of the subject loan, the borrower had already purchased and was in the process of purchasing additional undisclosed properties, which Walzak opined were likely investment properties because they were located in a resort area in Florida. Through investigation into the address on the borrower's driver's license and information from Accurint, a database that provides property ownership information, Digital Risk determined that the borrower never occupied the property even though he indicated that it was his primary residence. See Trial Tr. 349:17–351:23. Finally, Walzak's notes include a comment that the appraisal value of $290,000 was not supported because one of the comparable properties Flagstar used could not be verified by Clear Capital's AVM. See Pl. Ex. 190. Clear Capital's appraisal priced the property at $260,000. Id. By contrast, Griggs opined that Walzak's opinions were not reliable because they were based on information from a salary engine and from Accurint; that a debt obtained after the loan should not count as an undisclosed debt; that other information supported the reasonableness of the borrower's income; and that the comparable properties used by Clear Capital were farther away from the subject property and older than the comps used by the Flagstar AVM. See id. at 1247:16–1255:14.

For the loan ending in 7283, Walzak found that the borrower had misrepresented his income and debt obligations—opening two mortgages less than thirty days after the subject loan closed—and that Flagstar's original appraisal was unacceptable. The borrower claimed in his application that he was an operations manager at a health care company, but the borrower's credit report (obtained at the time of origination) reported that the borrower was self-employed, and Digital Risk, in verifying his employment with the health care company, discovered that the company had no record of the borrower's employment. As to the appraisal, Walzak found that Flagstar's initial $450,000 appraisal was not supported because the comparable properties used by Flagstar were determined to be dissimilar to the subject property and were of greater value, and the appraiser's attempted price adjustments were insufficient. See Trial Tr. 352:1–354:21. Griggs did not review this loan.

On the loan ending in 8376, Walzak determined that the borrower had misrepresented the occupancy of the subject property. The borrower claimed that the property was his primary home, although he owned a second condominium in the same Florida development. Digital Risk looked at the borrower's reported address on his driver's license, as well as the fact that the borrower never claimed Florida's homestead tax exemption (an exemption allowed for a homeowner's primary home) on the subject property, and determined that the borrower never occupied the property as his primary home. See Trial Tr. 355:4–356:6. Griggs found that this information did not conclusively prove a misrepresentation of occupancy because other post-closing sources reported that the borrower has treated both the subject property and the other property as his primary residence. Def. Ex. BBD 39 at 13.

For the loan ending in 9094, Walzak found that the loan file was missing documentation—including a VVOE form and letters of explanation regarding credit inquiries—and that the borrower had misrepresented his income and his intended occupancy of the property. While the borrower represented that he was the head golf professional at a Michigan golf course making a salaried income of $5,120 per month, Digital Risk verified the he was in fact paid hourly at $14.42 per hour and made $517 per month in commissions and bonuses, for a total of just over $3,000 per month. Walzak recalculated the DTI ratio for this borrower as 73.54%, which was significantly higher than the Flagstar guidelines allowed. As to the occupancy of the property, the borrower represented that the property, located in Florida, was his primary residence, but Digital Risk found that he never occupied the property and remains employed full-time in Michigan. See Trial Tr. 356:6–358:2. Griggs did not review this loan.

On the loan ending in 4808, Walzak found that the file was missing documentation, including any verification of employment, and that the borrower had misrepresented his debt obligations. In one of the more egregious failures to disclose debts, the borrower failed to inform Flagstar that he had an additional $7,000 per month combined mortgage payment on a $1 million loan and a $282,500 second mortgage, leading to a recalculated DTI ratio of 89.27%. See Trial Tr. 361:2–20. This loan was one of the three that Griggs recommended that Flagstar repurchase, as he found that the undisclosed debts meant that the borrower should not have qualified for a loan. See id. at 1246:7–21, 1282:23–1284:20.

As to the loan ending in 9046, Walzak found that the loan file was missing documentation, including information on the borrowers' income, insurance, and a final signed application, and contained a blank right of rescission notice. Additionally, Walzak found that the borrowers had misrepresented their occupancy of the property, as the borrowers worked in California but indicated that the subject property, located in Florida, was their primary residence, without explanation in the file as to how this could be so. See Trial Tr. 363:15–364:2. Griggs testified that additional information from Accurint showed that, shortly after the loan closed, the property was associated with a female who, he speculated, may have been a daughter of one of the co-borrowers, since she had the same last name and a reasonable age difference from the borrowers, suggesting that the borrowers may not in fact have entirely misrepresented the property's occupancy. See id. at 1232:10–1236:4.

In order to assess these and other sharp disagreements between Walzak and Griggs, the Court, before issuing this decision, conducted its own review of many of the loan files in evidence. Most of this review was conducted during the evening after the close of a trial day and therefore is only indirectly referenced, if at all, in the trial record. However, a record of one example of the Court's review exists with respect to the loan ending in 4977, which was initially brought to the Court's attentionduring defense counsel's cross-examination of Walzak. The borrower for this loan claimed to be the president of Regional Financial Group, located in Dearborn Heights, Michigan, with ten years in the industry and five years as president of the company, earning $16,790 per month. The Digital Risk file for this loan indicated that this income was unreasonable, as it reported that payscale.com showed the ninetieth percentile for a “president” of a company in Dearborn Heights as $9,020 per month. At trial, a live demonstration of payscale.com indicated that the ninetieth percentile income for a “president” in the financial services industry in Dearborn Heights would make $22,500. However, looking further into the loan file, Walzak noted that the borrower's credit report showed that the borrower was a manager-baker at a Dunkin' Donuts in 1995 and that, between at least 1997 and the time the loan was made, he was employed by the Detroit Police Department. Digital Risk verified with the Work Number that the borrower had been employed as a police sergeant starting in the mid–1990s. Walzak opined that, given this disparity, Flagstar should have obtained a VVOE or a letter of explanation. See Trial Tr. 636:24–639:12, 667:20, 693:9–705:5. On any fair inspection, it seems also that the borrower misrepresented his employment, if only by omission of his full-time position at the Detroit Police Department, and it seems equally clear that the borrower grossly overstated his reported income as, allegedly, a part-time president of a financial services company. Griggs did not review this loan and rated it “NR” as a life event loan, since the borrower later sent in a hardship letter to Flagstar, as the loan servicer states, that he was unable to make his payments because he was laid off from his positions both as a police officer and as a “broker.” See id. at 1274:12–1278:8; Def. Ex. BBB. Scott, Flagstar's head of underwriting, further confirmed that if Flagstar's underwriter had not confirmed the borrower's employment at the time of origination, that would have been a violation of Flagstar's underwriting guidelines. Trial Tr. 885:1–9.

Returning to other sample loans, it is apparent that Digital Risk made a mistake in its income testing for at least one borrower. For loan 4507, the borrower claimed that he was the owner of a business with ten years' experience earning $10,016 per month, and Digital Risk found a ninetieth percentile income in his geographic area of $8,058 per month. However, the Digital Risk file for this loan indicates that Digital Risk entered into the payscale.com system 3.17 years' experience—the value the borrower stated for the number of years at his prior address, rather than the ten years' experience the borrower reported. When ten years' experience was substituted during a live trial demonstration, the ninetieth percentile came out to $19,166 per month. See Trial Tr. 647:2–648:19, 674:21.

Finally, one of the loans Walzak withdrew from her tally of materially defective loans, the loan ending in 2331, indicates that Walzak did in fact look at compensating factors. The borrower reported in his loan application that he was self-employed in the drywall business for five years earning $16,894 per month, and Walzak found in the loan file neither a VVOE form nor any indication that the Flagstar underwriter conducted a reasonableness analysis when underwriting this loan. Payscale.com reported that the ninetieth percentile for this position was $12,547. However, Walzak withdrew her adverse finding on this loan because she determined that the borrower's liquid assets and credit history compensated for some of the concerns about his income, and she held out the possibility that his income might be higher than reported on payscale.com. See Trial Tr. 609:8–613:9.

In recounting the disputed expert testimony mention should also be made of Flagstar's proffered expert on the design and implementation of underwriting and due diligence reviews, consultant Jeffrey Nielsen,¹⁵ who opined at trial as to the quality of Walzak's review both in its own right and in comparison with the pre-transaction due diligence processes conducted by Clayton and Bohan. Trial Tr. 944:20–945:8, 948:22–25, 951:14–18. Nielsen testified to what he viewed as the characteristics of a quality review: (1) employment of individuals with strong subject matter expertise, training, and discipline in creating their work papers; (2) clarity of instructions; (3) objective criteria applied consistently; and (4) transparency of the process and the resulting work product so that the review's ultimate conclusions can be independently assessed and validated. Id. at 946:15–948:12. Nielsen's ultimate conclusion was that, given the flaws in the design and execution of Walzak's review and what Nielsen found to be a lack of professional quality, Walzak's findings of pervasive material and adverse breaches of the representations and warranties were fundamentally flawed and collectively unreliable. Id. at 952:10–13.

15 Nielsen holds a bachelor's degree in business administration and is the managing director of the financial services disputes and investigations service line at Navigant Consulting, Inc. Trial Tr. 942:21–943:18. Nielsen has spent his entire career at Navigant and was not qualified as an expert in underwriting per se. See id. at 1066:20, 1085:5–6.

In coming to his conclusion, Nielsen reviewed Walzak's deposition testimony and her spreadsheets summarizing each loan, and relied on his experience in designing and executing similar reviews. Trial Tr. 982:24–983:3. As to the 106–question spreadsheets, Nielsen found that the questions were vague and overlapping so that it was difficult to identify the actual substantive issues claimed to be present in the loan files. Id. at 984:4–25, 988:22–989:1. In addition to the issue of the confusing double-negative question U–126, Nielsen pointed to the obvious flaws in the wording of question U–113—“Did the underwriter obtain any critical missing or invalid data necessary to validate the overall probability of repayment?”—and noted that obtaining “invalid data” is a nonsensical proposition. Id. at 987:1–5; Pl. Exs. 189, 190.

Nielsen also found that Walzak's adverse findings in a variety of categories were inflated by derivative findings, such that it was difficult to determine what the foundational allegation was. Trial Tr. 992:1–6. For example, if Walzak found that a borrower's income was overstated, she also deemed DTI to be incorrect, leading to two adverse findings derived from the same underlying issue. Id. at 1104:10. However, Nielsen acknowledged that it was not unreliable to treat income and DTI as separate inquires, as DTI could be inaccurate for other reasons (undisclosed debts, arithmetic error) unrelated to income. Id. at 1107:4.

Third, Nielsen opined that the spreadsheets were opaque and failed to provide a foundation against which he could independently assess and validate or refute Walzak's findings. Trial Tr. 1007:1–5. Nielsen pointed to the loan ending in 9550 as an example in which Walzak's reviewer made no adverse finding in any of the specific questions on the spreadsheet, but Walzak reached an adverse conclusion based on the general, summary questions at the end of the spreadsheet, including question U126 on materiality, making it unclear on what that general finding was based. Id. at 994:9–16, 999:10–12. This inconsistency appeared to stem from the fact that Digital Risk at times found issues different from those identified by Walzak's underwriting team; Digital Risk's findings were then reported in the comments, while the underwriters answered the specific questions. See id. at 1113:6–15. (To the extent, however, that Nielsen complained that Walzak's summary questions themselves did not make clear what the specific identified issues were, Nielsen acknowledged that the “Event Level” definitions used in the due diligence reviews provide no greater degree of transparency or specificity. Id. at 1112:12.) As to Walzak's fraud findings, Nielsen testified that he was unable to replicate Walzak's fraud rate findings, either by tallying up those loans that had adverse findings in the “fraud” section of questions on the spreadsheet (amounting to about one percent of the loans) or by counting the comments that alleged a fraud-based claim. Id. at 1009:18–1010:6.

Finally, Nielsen concluded that Walzak failed to apply any unifying quality control standards to ensure consistency in her team of underwriters' responses. Trial Tr. 1056:6–22. Of the 120 files for which Walzak alleged that the borrowers' claimed incomes were unreasonable, Nielsen found that, in as many as 90 cases, Walzak's underwriters' answers on reasonableness of income conflicted with Digital Risk's conclusions (based on an online salary engine) in the comment fields for the same loans. Id. at 1055:21–1056:22, 1113:6–15. Similarly, Nielsen testified that he found Walzak's results inconsistent, such as where, at various times, missing VVOE forms were deemed material or immaterial, without a clear indication as to Walzak's reasoning for her differing conclusions (although Nielsen admitted that he could not say whether this difference was inappropriate without looking at the loan files, which he did not do as part of his review). Id. at 1121:17–1122:2. Nielsen further testified that he saw no evidence of Walzak's vetting her review process to address such inconsistencies. Id. at 1056:18–22.

Nielsen next compared Walzak's review to the due diligence review conducted by Clayton and Bohan in 2005 and 2006. To do so, Nielsen reviewed the documentation from the due diligence process and reconstructed the review to assess the standards and frameworks that Clayton and Bohan applied. Trial Tr. 1012:17–1013:11, 1017:22–24. According to Nielsen, both Walzak's review and the due diligence reviews were intended to provide findings that could be extrapolated pool-wide, as both reviews used random samples, and at least one of the ultimate questions was similar—whether Flagstar had materially complied with its underwriting guidelines in approving the loans involved in the transactions. See id. at 1019:4–24, 1088:5–6. One significant difference, according to Nielsen, is that Clayton and Bohan performed their reviews contemporaneously and were able to engage in an iterative process with Flagstar, in which the due diligence firms were able to remediate issues discovered in their reviews, including issues of missing documentation. Id. at 1030:8–1031:2. By contrast, Nielsen testified that Walzak's review was effectively a “day one” process, that is, the first round of the due diligence process without any feedback from Flagstar. Id. at 1035:6–12. Indeed, Walzak acknowledged that she was unable, in the context of litigation, to seek from Flagstar any clarification of the guidelines or additional documentation that was missing from the file. Id. at 572:17–573:24.

Nielsen noted that nineteen loans were reviewed as part of both a due diligence review and Walzak's review. See id. at 1045:6–12. While all nineteen eventually passed the due diligence review, Walzak found seventeen to be materially defective. Id. Among these seventeen with conflicting findings, Nielsen testified that, for some loans, the due diligence providers deemed the borrower's stated income supportable, while Walzak concluded that the income was unreasonable, and, for two loans, issues of missing documentation were remediated through the due diligence process, moving them from a grade of Event Level 3 to Event Level 1. See id. at 1048:8–1049:5, 1051:12–17.

Finally, Dr. Joseph Mason, Assured's damages expert,¹⁶ testified at trial that he constructed “an exact replica of the securitization Trusts” that allowed him to model “the cash flow into [the] Trusts” in order to simulate what would have happened had Flagstar repurchased the proportion of loans estimated to be defective in the two Trusts. Trial Tr. 421:8–10, 427:8–10. In creating this model, Dr. Mason relied on the Transaction Documents to model the structure of the Trusts, the Trustee reports for data on actual cash flows, and Walzak's defective loan findings. Id. at 427:15–16, 433:23, 446:7. Dr. Mason described his model as “conservative,” as he based his calculation only on loans that had defaulted or were more than 120 days delinquent (the “charged-off” loans), even though the contractual “repurchase or cure” remedy applies to all defective loans regardless of payment status. Id. at 429:9–10, 454:7–9. Dr. Mason also calculated the percentage of charged-off loans Flagstar was obligated to repurchase in terms of the dollar value of the loans, rather than a straight count of loans; thus, he divided the original loan balance of defective, charged-off loans by the total original loan balance of all charged-off loans and concluded that Flagstar should have repurchased 88% of the charged-off loans in the 2005–1 sample and 76% of the charged-off loans in the 2006–2 sample. Id. at 428:19–25. To simulate how the defective loans affected the cash flows into the Trusts, Dr. Mason then extrapolated those rates to the overall pools by multiplying the actual losses that the Trusts suffered by the percentage of defective, charged-off loans. Id. at 429:18–430:4.

16 Dr. Mason is the Hermann Moyse, Jr./Louisiana Bankers Association Endowed Professor of Banking at Louisiana State University and a senior fellow at the Wharton School of Business at the University of Pennsylvania. Trial Tr. 424:14–17. Dr. Mason holds a doctorate in economics. Id. at 424:5.

Dr. Mason concluded that, if Flagstar had begun repurchasing defective, charged-off loans in January 2009 (the date of Assured's first repurchase demands), there would have been sufficient cash coming into the Trusts to meet all the Trusts' payments obligations to the bondholders, and Assured either would not have paid any claims or would have been fully reimbursed for all claims paid. Trial Tr. 428:4–16. Indeed, Dr. Mason further determined that if Flagstar had repurchased the defective loans, there now would be excess collateralization of $20.5 million in the 2005–1 Trust and $21 million in the 2006–2 Trust that would insulate Assured against future claims. Id. at 432:10–13. Dr. Mason further testified that the Trusts would have absorbed all of the losses and thus Assured would have been fully reimbursed under a wide range of defect rates; that is, even if Walzak were incorrect in her findings of material breach on some of the loans in the sample, Flagstar would still be obligated to reimburse Assured for the entirety of the claims it has paid. Id. at 429:14–15. By Dr. Mason's calculation, as of the date of trial, Assured's total damages amounted to $111 million, inclusive of interest but exclusive of costs and attorney's fees. Id. at 422:16–21, 430:13–19.

Dr. Mason testified that he based his damages calculation on the assumption that Flagstar's repurchase obligations became effective as of January 26, 2009, the date of Assured's first repurchase demands for both Trusts, at which point interest began to run on the claims Assured paid to bondholders. See Trial Tr. 429:23–25. Dr. Mason testified that he applied the contractual interest rate specified in the I & Is (prime rate plus two percent) from the date Assured paid a claim until Assured should have been reimbursed, see Pl. Ex. 90 App. I at 2, and then applied statutory interest (nine percent under New York law) if Assured never should have paid a particular claim or from the date it should have been reimbursed up to September 25, 2012. Id. at 431:1–18.

In a different damages scenario about which he testified on cross-examination, Dr. Mason assumed that Flagstar was aware of widespread breaches of the representations and warranties as of the offering date of the securities. Trial Tr. 441:15–17. In this scenario, Dr. Mason testified that he calculated the cash that should have flowed into the Trusts from Flagstar's repurchases of defective loans by multiplying the “underwriting defect rate” (the original dollar value of defectively underwritten loans divided by the original pool balance) by the actual gross loss experienced by the Trusts each month. Id. at 436:14–25. Assuming this altered cash flow, Dr. Mason subtracted what Assured should have paid from what it actually paid, and then determined the net present value of that difference by adding to those amounts statutory interest, reasoning that the contractual interest rate would be inapplicable if Flagstar was responsible for repurchasing defective loans as of the offering date of the security and thus Assured would never have paid any claims to which the contractual rate would apply. Id. at 439:22–23, 441:15–443:10; see also Def. Exs. AYB, AYG. This model came out to a similar damages figure.

On cross-examination, Dr. Mason conceded a few errors in his calculations of the damages owed to Assured. First, Dr. Mason acknowledged that he failed to subtract from the amounts owed to Assured those claim payments relating to the 2005–1 transaction for which Flagstar had already reimbursed Assured. Specifically, Assured received reimbursements of $485,323.03 on August 25, 2010, $400,337.63 on January 25, 2011, and $62,678.60 on November 25, 2011, amounts which are not reflected in Dr. Mason's spreadsheet calculating the damages owed to Assured. Trial Tr. 446:20–447:24. Second, Dr. Mason included in his count of charged-off loans some loans that were less than ninety days delinquent and thus should not have been included in the total. Id. at 460:4–6, 465:2–3, 466:1–2. For example, eleven loans in the 2006–2 sample were not charged off as of January 26, 2009, but were nevertheless counted in the total of charged-off loans and defective, charged-off loans, where applicable. Id. at 463:12–16.

RESOLUTION OF THE CONFLICTING EXPERT TESTIMONY

The Court finds that the evidence proffered by both Dr. Lipshutz and Dr. Mason is clear, credible, and convincing. As more fully detailed below, the sample size suggested by Dr. Lipshutz provided an adequate basis for assessing whether the Trusts as a whole complied with or breached Flagstar's representations and warranties, and Dr. Mason provided a solid basis for calculating the damages occasioned by any such breaches. But the necessary connection between their two testimonies was the much more problematic testimony of Ms. Walzak as to what breaches occurred. In the end, however, the Court concludes that Ms. Walzak's testimony, though flawed in certain respects, was sufficientlyconvincing as to carry plaintiff's burden.

As a preliminary matter, it should be noted that Flagstar, at the close of Assured's case and again at the close of all the evidence, moved to exclude the expert testimony of Dr. Lipshutz, Ms. Walzak, and Dr. Mason under Federal Rule of Evidence 702. Since the Court reserved decision on the motions until at both points, they will now be considered both on the record as it existed at the close of Assured's case and on the record as it existed at the close of trial. See Trial Tr. 793:5–19.

In order for expert testimony to be admitted, Rule 702 requires that an expert's “specialized knowledge will help the trier of fact to understand the evidence or to determine a fact in issue,” that “the testimony is based on sufficient facts or data” and “is the product of reliable principles and methods,” and that “the expert has reliably applied the principles and methods to the facts of the case.” Fed.R.Evid. 702. In determining the admissibility of an expert witness's testimony, a court must “undertake a rigorous examination of the facts on which the expert relies, the method by which the expert draws an opinion from those facts, and how the expert applies the facts and methods to the case at hand.” Amorgianos v. Nat'l R.R. Passenger Corp., 303 F.3d 256, 267 (2d Cir.2002). However, “in accordance with the liberal admissibility standards of the Federal Rules of Evidence, only serious flaws in reasoning or methodology will warrant exclusion.” In re Fosamax Prods. Liab. Litig., 645 F.Supp.2d 164, 173 (S.D.N.Y.2009). This is especially true in the context of a bench trial, where “there is no possibility of prejudice, and no need to protect the factfinder from being overawed by ‘expert’ analysis.” Victoria's Secret Stores Brand Mgmt., Inc. v. Sexy Hair Concepts, LLC, 07 Civ. 5804, 2009 WL 959775, at *8 n. 4 (S.D.N.Y. Apr. 8, 2009). Particularly in a bench trial, “[v]igorous cross-examination, presentation of contrary evidence, and careful ... [attention to] the burden of proof are the traditional and appropriate means of attacking shaky but admissible evidence.” Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579, 596, 113 S.Ct. 2786, 125 L.Ed.2d 469 (1993).

As to Dr. Lipshutz, Assured's sampling expert, Flagstar challenges neither his qualifications as an expert nor his fundamental methodology in drawing the 400–loan random samples from the 2005–1 and 2006–2 loan pools. However, Flagstar raises two challenges to the design of Dr. Lipshutz's samples as applied to certain issues in this case.

First, addressing the interplay between Dr. Lipshutz's testimony and Dr. Mason's calculations, Flagstar argues that Dr. Lipshutz failed to test the representativeness of the loan samples against certain variables—loan payment status and original principal balance—that are particularly relevant to Dr. Mason's damages calculation. See Trial Tr. 261:2–8. Thus, Flagstar claims it would be inappropriate for the Court to rely on Dr. Lipshutz's samples to extrapolate those damage calculations to the entirety of the loan pools. However, as Dr. Lipshutz testified at trial, the purpose of using a pure random sample of sufficient size is so that one can estimate the proportion of loans in the overall loan pools exhibiting any characteristic, not just those tested. Id. at 252:14–24, 257:24–258:4. The Court agrees. Moreover, while the burden is on Assured, Flagstar presented no evidence at trial suggesting that the samples are not representative as to these particular variables. Overall, the Court concludes that the samples are duly representative of the kind of loan characteristics used in Dr. Mason's calculations.

Second, Flagstar argues that Dr. Lipshutz calculated the minimum representative sample sizes for the two transactions—371 loans for the 2005–1 pool, and 358 for the 2006–2 pool, see Trial Tr. 254:3–19—based on a mistaken understanding of the inquiry at hand, so the samples were incorrectly designed. Dr. Lipshutz determined what sample size would be required with the understanding that the samples would be used “to make a binary decision on each loan,” that is, whether the loan file conforms to the representations and warranties made in the Transaction Documents. Id. at 257:8–12. Flagstar argues that the key determination of “materiality” is not a binary determination, but rather a question of degree of severity, which the sample is not designed to accommodate. However, the ultimate determination that Walzak made—and that the Court must make—is in fact a binary decision: did any given loan in the sample breach the representations and warranties in a way that was material and adverse to Assured's interests, or not. Although there are several bases on which a loan may be held not to materially violate Flagstar's representations and warranties, and assessment of those bases may involve questions of the severity and number of underwriting errors and misrepresentations, nevertheless, unless all those bases are met, Flagstar is not liable. In other words, the fundamental decision is a binary one: whether or not Flagstar was obligated to repurchase any given loan in the sample. Thus, Dr. Lipshutz did not err, Flagstar's challenge to the representativeness of the samples is rejected, and Flagstar's motion to exclude Dr. Lipshutz's expert testimony is denied.

Jumping ahead to Dr. Mason, Flagstar argues that, even putting aside his necessary reliance on Ms. Walzak's conclusions, discussed below, his damages calculations are unreliable because they are based on too small of a sample to be extrapolated pool-wide with a sufficient level of confidence in the result. Dr. Mason testified that he calculated damages, in one model, based only on loans that had been charged off, amounting to no more than 47 out of 400 loans in the 2005–1 sample and 37 out of 400 loans in the 2006–2 sample. See Def. Exs. AYM, BBK. From there, he determined what proportion of the charged-off loans in the sample were also defectively underwritten according to Walzak's findings. See Trial Tr. 428:19–22. Dr. Mason then used that percentage to calculate what percentage of the overall losses in the securitization was due to defective loans. See id. at 429:18–21. Flagstar's counsel raised with Dr. Lipshutz on cross-examination whether creating a stratified sample—one in which a random sample is taken from a sub-population rather than an overall population—would be required to accurately extrapolate Assured's damages calculation to the entire population of loans. In particular, defense counsel raised the concern that “if one does not stratify, then the number of charged-off loans found in the random sample may be too small to obtain a sufficiently accurate measure of the portion of breaches found in the subpopulation of charged-off loans to the entire transactions.” Id. at 282:14–18. In response, Dr. Lipshutz, while contending that the best estimate of a subset of the population of loans would remain the estimate obtained in the sample, confirmed that the variance of the interval—the size of the confidence band with which the estimate can be projected—may be affected by a smaller sample size. Id. at 282:19–20, 283:10–19.

The Court is not persuaded that stratification is required, essentially for the reasons given by Dr. Lipshutz. But even assuming that Flagstar's argument that stratification would be required were correct, the Court finds that Dr. Mason's testimony is sufficiently reliable nonetheless, as Dr. Mason testified that Assured would be reimbursed in full under a wide range of defect rates. Trial Tr. 429:14–15. Thus, Flagstar's challenge to the admissibility of Dr. Mason's testimony is denied.

The most contested testimony in this case, however, was that of Ms. Walzak, without which the opinions of Dr. Lipshutz and Dr. Mason would be irrelevant; and Flagstar centers its most pointed challenges to admissibility on her testimony.

Flagstar first claims that Walzak fails to qualify as an expert, both because she is not a neutral expert and because her experience does not qualify her as an expert in mortgage underwriting. Flagstar notes that Walzak's curriculum vitae presents her as a “results-oriented” management consultant, not an objective expert in underwriting, and points out that she has made a substantial income in the recent past testifying on behalf of mono-line insurers in cases like this one. See Def. Ex. AZO; Trial Tr. 477:24–478:1, 387:6, 392:3. However, while Walzak's description of herself as “results-oriented” exhibits a certain lack of professional restraint, it also candidly reflects what every judge knows: that most party-hired experts are hired in part because of this known predisposition. Such biases must be taken into account in evaluating expert testimony, but, except in extreme cases, they are not a basis for excluding experts' testimony altogether. Indeed, Flagstar itself hired an expert, Jeffrey Nielsen, from Navigant Consulting, whose website touts the company's ability to help clients “protect value” “in the face of a critical business risk” and promises to “deliver[ ] results”—statements that are not that different from the way in which Walzak portrays herself. See Pl. Exs. 510, 511; Trial Tr. 1069:19–1070:19.

As to Walzak's qualifying experience, it is true that she served as a front-line underwriter only at the beginning of her career and has more recently moved into quality control and risk management consulting. See Def. Ex. AZO; Trial Tr. 481:1–6. But that is to be expected of any successful expert, and it is hardly a disqualification. Indeed, John Griggs and Marni Scott, Flagstar's underwriting expert and head of underwriting, respectively, face the same problem: as these witnesses have gained experience in underwriting, they have moved into related areas, like quality control and management, and away from direct responsibility for underwriting loans. See Trial Tr. 864:5–15 (describing Scott's managerial role); 1143:8–25 (describing Griggs' role as a management consultant). Moreover, Walzak's experience, as reported in her curriculum vitae, indicates that she has remained intimately familiar with mortgage underwriting practices in the recent past, so as to qualify her as an expert able to testify to industry standards during the relevant time period. See Def. Ex. AZO at 4 (listing recent affiliation with, inter alia, the Mortgage Bankers Association of America).

Flagstar next challenges Walzak's testimony on the ground that she failed to define when a breach of a representation or warranty “materially increases the risk profile of a loan,” and thus failed to employ an articulable, consistent methodology in arriving at her expert opinion. When asked when a file might be deficient on its face, Walzak failed to articulate a clear standard, and when asked why she withdrew four loans before trial, Walzak was unable to give a clear reason as to why she deemed the breaches immaterial, which Flagstar asserts indicates a lack of clear and articulable standards and a reason to lack confidence in her results. Trial Tr. 581:24–582:15; 608:7–12. Thus, Flagstar claims, what the Court is being asked to do is accept that Walzak's subjective judgment and anecdotal descriptions of a few dozen loans and merely trust her judgment as to the remaining loans in the sample. Flagstar argues that this is not the kind of reliable, verifiable methodology required of expert testimony under Rule 702, and it instead amounts to Walzak's mere “ ipse dixit.” See Kumho Tire Co., Ltd. v. Carmichael, 526 U.S. 137, 157, 119 S.Ct. 1167, 143 L.Ed.2d 238 (1999).

But the absence of mechanical standards for defining all the circumstances that might create a material breach of Flagstar's representations and warranties with respect to any given loan reflects not a failure of methodology, but a candid recognition of the multi-variable nature of the inquiry. Indeed, Walzak, Nielsen, and Griggs all testified at trial that the question of whether a breach of the representations and warranties is material is a function of all of the information and circumstances presented in each loan file. See Trial Tr. 555:25–556:9 (Walzak); 1044:18–24 (Nielsen); 1171:17–1174:17 (Griggs). Not every violation of Flagstar's underwriting guidelines constitutes a material breach; rather, a substantial violation of an individual underwriting guideline may be material, or many smaller breaches may cumulatively increase the risk profile of the loan to a sufficient degree. Id. at 332:6–17. Moreover, where the guidelines allow for the consideration of compensating factors, issues that might otherwise be material might be mitigated by the presence of a particularly favorable status on another factor. Id. at 532:24–533:23; 1172:8–13.

Inevitably, this means that the opinion of any expert testifying on this issue involves a degree of subjectivity. But this is not a basis for rejecting such an opinion on the ground that the methodology is “unreliable.” Indeed, the testimony at trial indicated that the methodology applied by Ms. Walzak is the same kind of methodology underwriters apply in the field. It is the product of experience, and in this case, can be (and was) adequately tested by cross-examination.

As the Supreme Court explained in Kumho Tire, experts qualified by their experience may testify to their conclusions as long as they exhibit “in the courtroom the same level of intellectual rigor that characterizes the practice of an expert in the relevant field.” 526 U.S. at 152, 119 S.Ct. 1167. Walzak articulated the sources of information on which she relied—the information in the loan file, Flagstar's guidelines, and various tools like online salary engines and AVMs—and applied her experience in the underwriting industry to the resultant findings to make a determination as to whether the deficiencies in a particular loan rose to the level of materiality. See Lippe v. Bairnco Corp., 96 Civ. 7600, 2002 WL 15630, at *2 (S.D.N.Y. Jan. 7, 2002) (allowing experts to testify to “the customs and standards of an industry” and “how a party's conduct measured up against such standards”).

Even if, as the witnesses testified, each loan is unique, the collection of deficiencies Walzak found and her methods of recognizing those problems were not: to take a few examples, Walzak's reviewers looked to inquiries on a borrower's credit report to determine whether a borrower had failed to disclose a debt obligation; Digital Risk looked to online salary engines, employer verification, and/or the Bureau of Labor Statistics' data to determine whether a borrower's stated income was reasonable; and Walzak's underwriters looked in the loan file to see if all required documentation had been obtained. Trial Tr. 331:22–332:4, 350:9–11, 602:19–603:22, 677:5–7. Under the circumstances, this was an adequate methodology to allow Walzak to obtain the evidence of breach necessary to make a determination, based on her experience as an underwriter, as to whether those breaches increased the risk to Assured on a given loan.¹⁷

17 It is true that, as testified to by Scott, Walzak misinterpreted the Flagstar's guidelines in one respect, because, contrary to her belief, Flagstar's guidelines did not require a letter of explanation on credit inquiries on the loans at issue in this case. Trial Tr. 915:6–13. But this error did not sufficiently affect her analysis in the vast majority of loans; and the Court, in any case, has undertaken to exclude from its factual determination the few instances in which this error was significant.

Moreover, in loan after loan, Walzak detailed the ways in which the loan file failed to meet Flagstar's guidelines, and Flagstar was unable to mount a persuasive challenge to her conclusions in the vast majority of cases. Indeed, the Court's own review of various loan files convinced the Court that Ms. Walzak's views were right on target, for many of the loans (including several described above) exhibited precisely the defects—many of which were blatant—that Ms. Walzak posited. In short, the Court finds Ms. Walzak's methodology not only appropriate to the courtroom but corroborated by the Court's own review.

Next, Flagstar challenges the tools on which Walzak relied in deriving her conclusions. In particular, Flagstar argues that online salary engines, which Walzak used as the major tool in determining whether a borrower's stated income was “reasonable,” are unreliable because they are based on voluntarily submitted information and because they reflect only current salary data, which may not accurately represent salary information from five years ago and before the economic downturn. Trial Tr. 1187:7–19. Similarly, Flagstar's witness, Griggs, criticized Walzak's use of Accurint and Clear Capital's AVM, which he argued were inherently unreliable sources of information. Id. at 1230:12–23, 1235:1820.

As an initial matter, however, it appears that these tools are commonly relied upon in the field of underwriting. Flagstar itself recommends the use of salary.com to determine reasonableness of income, see id. at 321:17–24; Griggs' own team looked to Accurint when challenging some of Walzak's allegations, id. at 1232:10–1236:4; and Scott testified that Flagstar trained its underwriters to use similar online tools, id. at 911:1–16, and that Flagstar regularly validated appraisals using a different AVM, id. at 909:10–21. As to the online salary engines specifically, the Court is satisfied that any concerns about reductions in average income in the intervening half-decade since the origination of the loans is mitigated by the fact that Walzak only treated a borrower's stated income as unreasonable when it far exceeded an online salary engine's reported ninetieth percentile for individuals in the borrower's field and geographic area. Thus, the Court finds these tools to be sufficiently reliable so as to appropriately underlie Walzak's analysis of the loan samples.

Flagstar also argues that Walzak's survey of industry underwriters, through which she sought to determine the parameters of the industry consensus for various “reasonableness” determinations in underwriting, was inherently flawed and cannot constitute an “industry standard” on which reliable findings can be based. See Trial Tr. 505:6–508:9. Walzak testified that she gave this review to her underwriters to provide guidance as to what should be considered “reasonable” where Flagstar's underwriting guidelines give leeway to its individual underwriters. Id. at 407:12–14, 505:3. A review of the survey indicates, however, that for some issues, no clear consensus exists to suggest an “industry standard” that Walzak's team could apply. See id. at 505:6–8. But the Court finds that this issue appears at best tangential to the work that Walzak's reviewers did. Because, moreover, Walzak set clear standards herself as to some of these issues—deciding, for example, to use the ninetieth percentile income as the bar for reasonable income—and because Walzak made the ultimate determinations of materiality of breach based on her own industry experience as an underwriter, there was no material error here.

Flagstar next claims that Walzak's review process lacked internal controls and quality control, and therefore Walzak failed to “reliably appl[y] the principles and methods to the facts of the case,” as is required by Rule 702. As described above, Walzak's oversight of her team of underwriters was rather limited, and this led to results that at times appear inconsistent. Perhaps most significantly, Nielsen testified that he was unable to reconstruct the fraud rates claimed by Walzak, either by tallying the comments relating to fraud or the fraud section of questions filled out by Walzak's underwriters. Id. at 1009:1–1010:6.

But this misapprehends the use Ms. Walzak made of the data and comments supplied by her underwriters, as well as Digital Risk and Clear Capital. These groups basically collected factual data and, sometimes, offered useful comments. While, where they found no problems, Walzak accepted their findings, when they did find problems, Walzak conducted an independent review and made the ultimate conclusions, Her final review, the Court finds, was essentially consistent, and any inconsistencies on the part of her supporting team were therefore immaterial.

Moreover, Walzak's testimony indicates that she remained available for questions from her reviewers throughout the process and reviewed all of their work as it came in, including an initial sample to test the consistency of the underwriters' understanding of the guidelines. Trial Tr. 489:2–9. As to Clear Capital, Walzak discussed at least with Digital Risk what Clear Capital would be doing, and Walzak received a demonstration of the product before the process began. See id. at 379:20–380:2, 685:1–8. Moreover, Flagstar's substantive challenge to the use of the Clear Capital AVM amounts to little more than a claim that the Clear Capital AVM was “not well known” in the industry, without any factual information suggesting that the model or the data underlying it were inadequate or insufficient. Id. at 914:13–14. And, again, Walzak reviewed each of Digital Risk's findings in making her ultimate determination of materiality. Id. at 311:24–312:2.

To be sure, it is clear from the testimony at trial that Digital Risk made occasional mistakes that Walzak failed to catch—such as the case of the loan described above for which Digital Risk entered the value of the borrower's years in his home as the borrower's years of experience in his employment. See Trial Tr. 647:2–648:19. However, given the volume of loans that Walzak reviewed, it seems inevitable that occasional errors would slip by, but there has been no showing that such errors were either frequent or material.

Finally, Flagstar challenges the credibility of Ms. Walzak's testimony that she reviewed each loan file in which her reviewers noted a potentially material issue and came to her own conclusions regarding the materiality of the breaches indicated by her reviewers and/or by Digital Risk. See Trial Tr. 310:21–25. In particular, Flagstar makes much of the fact that Walzak failed to mention this full loan-file review in her deposition—when she testifiedthat she looked at the loan file only if she saw inconsistencies in the spreadsheet or something sparked her interest, see id. at 495:2–20; that Walzak could not have had time to thoroughly review each loan file given the time constraints of a four-week review period and Walzak's other obligations, see id. at 583:19–593:22; and that Walzak could not remember a situation in which she disagreed with her reviewer's conclusions, see id. at 562:16–17.

This, however, is not a challenge to admissibility under Rule 702, but rather a challenge to the credibility of Ms. Walzak's testimony. Treating it as such, moreover, the Court finds Ms. Walzak's testimony fully credible in this regard. As an experienced expert in this area, Ms. Walzak knew well that such personal review was critical to the admissibility of her testimony and the weight it would be given, and four weeks was more than ample time to review the 610 loan files that remained after her team's initial screening. The Court itself was able to review several such files in a matter of minutes, aided by the same kind of flagging of problematic issues that Walzak's team identified.

Because the Court has rejected each of Flagstar's contentions as to Walzak's expert analysis, the Court denies Flagstar's motion to exclude Walzak as an expert under Rule 702.

Having denied Flagstar's challenges to Assured's witnesses, the Court turns briefly to Flagstar's motion for judgment on partial findings under Federal Rule of Civil Procedure 52(c), made at the close of Assured's case. As the Court noted at trial, Assured's initial failure to enter Flagstar's guidelines and all of the loan files that Walzak reviewed into evidence on Assured's case presented a potential problem of proof for Assured. However, Walzak testified to the substance of Flagstar's guidelines, and the Court concludes that this provided sufficient evidence to allow Assured to overcome the Rule 52(c) motion, especially in light of the fact that Assured's claims that Flagstar breached its contractual representations and warranties were not entirely dependent on a breach of Flagstar's guidelines. Additionally, it should be noted that Assured cured these evidentiary issues during Flagstar's presentation of its case by submitting into evidence the relevant Flagstar underwriting guidelines, see Pl. Ex. 96, as well as the loan files for the 126 loans reviewed by both Walzak and Griggs, see Def. Exs. BBB 1–126. The Court therefore denies Flagstar's Rule 52(c) motion as of the close of Flagstar's case as well.

The Court turns next to the heart of the case: whether Flagstar breached its contractual responsibilities by failing to repurchase materially defective loans from the securitization Trusts. The Transaction Documents are expressly governed by New York law. Pl. Ex. 90 § 6.05; Pl. Ex. 451 § 7.02; Pl. Ex. 198 § 8.02. Under New York law,

In order to recover from a defendant for breach of contract, a plaintiff must prove, by a preponderance of the evidence, (1) the existence of a contract between itself and that defendant; (2) performance of the plaintiff's obligations under the contract; (3) breach of the contract by that defendant; and (4) damages to the plaintiff caused by that defendant's breach.
Diesel Props S.r.l. v. Greystone Bus. Credit II LLC, 631 F.3d 42, 52 (2d Cir.2011). The parties do not contest that they entered into valid contracts or that Assured has fully paid the claims required to be paid under the Transaction Documents.

The first issue, then, is whether Flagstar breached the relevant contracts. Under the terms of the Transaction Documents, as described above, Flagstar is liable for any breach of the contractualrepresentations and warranties that “materially and adversely affects the interest of the Issuer, the Noteholders or the Note Insurer in the related Mortgage Loan,” regardless of whether Flagstar knew that the substance of the representation or warranty “was inaccurate at the time [it] was made.” Pl. Ex. 198 § 2.04(b); see also Pl. Exs. 198 § 8.06; 451 § 7.08 (making Assured a third-party beneficiary to the SSAs and MLPAs); See Pl. Ex. 90 § 2.03(h) (expressly incorporating the representations and warranties). Upon becoming aware of such breaches, Flagstar has an obligation to cure the deficiencies in the loans or repurchase the defective loans within 90 days. See Pl. Ex. 198 § 2.04(d). If Flagstar fails to do so, it has breached the terms of the contract, and Assured is entitled to enforce the “repurchase or cure” remedy contained in the Transaction Documents. See Pl. Ex. 451 § 3.02(c) (describing the cure-or-repurchase process as the “sole remedy” against Flagstar “for the breach of a representation or warranty with respect to a Mortgage Loan”); see also Pl. Ex. 198 § 2.04(e) (to similar effect). It is clear that Flagstar has repurchased only a small handful of loans over the course of these transactions, see Pl. Ex. 407 (listing Flagstar's repurchases); thus, the critical issues that remain are (1) whether Assured proved that the loans materially breached the contractual representations and warranties; (2) whether Flagstar was sufficiently aware of this fact to trigger the repurchase-or-cure remedy; and (3) what damages, if any, Flagstar owes to Assured.

Most of the trial focused on the question of whether the loans underlying the transactions materially breached the contractual representations and warranties. The representations and warranties that Assured claims Flagstar breached are that (1) “Each Mortgage Loan was originated in good faith and in accordance with [Flagstar's] underwriting guidelines,” and (2) “No error, omission, misrepresentation, negligence, fraud or similar occurrence with respect to a Mortgage Loan has taken place on the part of any person, including, without limitation, the Mortgagor, any appraiser, any builder or developer, or any other party involved in the origination of the Mortgage Loan or in the application of any insurance in relation to such Mortgage Loan.” Pl. Ex. 451 § 3.02(a)(36), (65). In its opinion denying Flagstar's summary judgment motion, the Court determined that the further requirement that breaches of these representations and warranties be “material and adverse” means that “plaintiff must only show that [Flagstar's] breaches [of the representations and warranties] materially increased its risk of loss.” Mem., 892 F.Supp.2d 596, 603 (S.D.N.Y.2012). As discussed above, whether breaches of the representations and warranties “materially increase” Assured's risk of loss is a function of all of the circumstances presented in each unique loan file. Trial Tr. 555:25–556:9 (Walzak); 1044:18–24 (Nielsen); 1171:17–1174:17 (Griggs).

One hard and fast rule that Walzak applied was that evidence of fraud—evidence that the borrower had affirmatively lied—was an automatic breach of the representations and warranties, regardless of whether there was also evidence of violations of Flagstar's guidelines. Trial Tr. 535:5–11. Further, as Brewer testified, Assured would have considered misrepresentations of borrower income, occupancy, and debt obligations material in its decision to insure the Flagstar transactions, as such misrepresentations indicate that a borrower may not be as creditworthy as Flagstar and Assured initially believed. Id. at 55:17–24, 59:1–7, 60:4–8. Even Griggs testified that when a borrower misrepresents his income or omits a debt obligation,that borrower's overall truthfulness is compromised and the risk of loss associated with that loan increases. Id. at 1284:20, 1286:21. The Court finds that this testimony well supports a finding that evidence of fraud is inherently material.

Additional principles help guide the determination of material breach. First, where exposure of a misrepresentation or a miscalculation of a relevant data point—e.g., income, assets, debts—would lead to a modified CLTV or DTI calculation that fell outside Flagstar's guidelines permissible range, the misrepresentation or miscalculation would be deemed a material breach if the deviation from Flagstar's standards was sufficiently large that no compensating factors could have made up for the deviation. It is possible to see this kind of determination in the two loans that even Griggs deemed to be materially deficient because they were approved without taking account of undisclosed, already existing debts that, once exposed, drove the DTI far above the permissible limit set by the Flagstar guidelines. See Trial Tr. 361:2–20, 1246:7–21, 1282:23–1284:20. Second, where significant documentation was missing from the loan file such that it was impossible to verify that the required checks were completed, Walzak deemed this a material failure to follow Flagstar's guidelines if no adequate compensating documentation existed in the file. This approach can be seen at work in the case of a loan in which no VVOE existed in the file to substantiate the borrower's claimed income, but other documentation suggested that the borrower's income might be reasonable, and Walzak determined the breach to be immaterial. Id. at 609:8–613:9.

Applying these principles, Walzak found 606 of the 800 loans reviewed to be materially defective, specifically, 85% of the 2005–1 sample and 66.5% of the 2006–2 sample. See Trial Tr. 315:2–10. The 606 defective loans fall into the following categories: 86 loans were made to borrowers who failed to disclose significant debt obligations (including loans that had closed after the closing of the subject loan), id. at 321:6–15; 131 loans were made to borrowers who stated their income at unreasonably high levels, id. at 322:4–10; 125 loans involved DTI limits, that once properly adjusted, exceeded Flagstar's guidelines, id. at 325:16–326:10; 24 loans were based on overvalued appraisals, id. at 326:23–327:17; 18 loans involved CLTV calculations that, once properly adjusted, exceeded Flagstar's guidelines, id. at 326:13–18; 17 loans were based on borrowers' misrepresenting their occupancy of the subject properties, id. at 328:15–16; 49 loans were made to borrowers who had insufficient cash reserves to meet Flagstar's requirements, id. at 324:22–325:12; 208 loans were made without the required VVOE form, id. at 323:6–7; and 179 loans were made without other required documentation, id. at 331:22–332:4.¹⁸ Even after eliminating the few instances where Walzak's finding of breach was based on her misinterpretation of Flagstar's guidelines relating to letters of inquiry (see above), a huge percentage of the loans in the two pools were defective.

18 The combined total of 837 exceeds the 606 defective loans because certain loans were defective in more than one category.

The Court finds that, with minor exceptions, Walzak's conclusions in these respects were fully credible and corroborated in numerous ways. To begin with, the representative instances described by the parties at trial (and illustrated in findings of fact above), were, for the most part, exactly as Walzak had categorized them. Indeed, the Court's own independent reviewof several of the loan files completely bore this out.

Furthermore, Flagstar, while repeatedly attacking Ms. Walzak's methodology, was notably unable to show actual instances where the loan files themselves did not contain material breaches of the guidelines. Thus, even though Flagstar's expert, Griggs, agreed with Walzak as to material deficiency on only three loans, see Trial Tr. 1246:2–16, he was unable to illustrate his conclusions with respect with respect to specific loan files in a manner that the Court found convincing; and for the most part he did not even try. Indeed, the filters Griggs employed—not reviewing the 484 loans that had paid in full or are still performing and the 43 “life event” loans, and deeming immaterial any breach of the representations and warranties if the borrower made his or her payments for twelve months—meant that Griggs failed to challenge the vast majority of Walzak's findings. Assured is free to prove the fact of Flagstar's breach, if not its damages from that breach, by demonstrative pervasive, materially defective underwriting equally for performing, delinquent and defaulted loans, as the “cure or repurchase” remedy in the Transaction Documents is not limited to defaulted or delinquent loans. Thus, the 484 loans that went unreviewed based on their payment status remain relevant to the Court's determination of material breach.

Moreover, given the Court's prior ruling that “the causation that must here be shown is that the alleged breaches caused plaintiff to incur an increased risk of loss,” Mem., 892 F.Supp.2d at 603, it is irrelevant to the Court's determination of material breach what Flagstar believes ultimately caused the loans to default, whether it is a life event or if the underwriting defects could be deemed “immaterial” based on twelve months of payment. Risk of loss can be realized or not; it is the fact that Assured faced a greater risk than was warranted that is at issue for the question of breach. Thus, the Court concludes that Walzak's findings were largely unrebutted by Griggs.

As for the 126 loans that Griggs did review, Griggs claimed that 80 did not involve material breaches, i.e., the 13 that Griggs rated a 1 and the 67 that Griggs rated a 2. Trial Tr. 1268:22–1269:21. However, Griggs testified that loans characterized as a 2 include those loans that met Griggs' twelve-month successful payment filter (which, as noted, the Court regards as improper), loans in which an undisclosed debt was discovered but for which Griggs believed that potential, but undocumented rental income should be considered in calculating DTI (an unsubstantiated assumption that the Court rejects), and loans for which Griggs found missing documentation issues to be “curable” based on Flagstar's supplementation of its production with documents found in other loan files. Id. at 1193:3–1200:6, 12:14:21–22, 1304:2–15. Taking Griggs' testimony that, absent the twelve-month filter, some of the category 2 loans would have been rated a 3, see id. at 1307:9–16, and disregarding Griggs' speculative assumption of potential rental income, the number of category 2 loans would be substantially reduced and category 3 loans equally increased, indicating that Walzak was indeed correct on many of her findings. That leaves, at minimum, the 13 category 1 loans on which Griggs and Walzak disagree. Even assuming Griggs is correct on all 13 of those loans he rated a 1, Walzak's findings largely remain uncontested on the vast majority of loans in the samples.

It should also be noted that Griggs' reviewers deemed only 30 allegations out of the 1280 that they reviewed—less than 3% of the allegations—to be “not clear.” See Trial Tr. 1169:22–23. This effectively contradictsNielsen's criticism of Walzak's questions as vague and unclear and her review as opaque.¹⁹ See id. at 988:6–13, 1007:1–5. Moreover, even if, as Griggs and Nielsen both testified, Walzak's individual adverse findings were inflated by derivative and overlapping findings, this inflation was largely irrelevant. See id. at 992:1–6. Walzak testified that she based her ultimate determination as to whether Flagstar materially breached the representations and warranties on the totality of the circumstances present in each loan file and the seriousness of Flagstar's errors, not by tallying the number of adverse findings per loan.

19 Nielsen was not tasked with making any determination as to whether the loans breached the representations and warranties, and made no such conclusions at trial.

The Court has considered Flagstar's other objections to Walzak's conclusions and finds them sufficiently without merit as to not warrant further discussion with one exception that merits brief mention, to wit, Flagstar's challenge to the use of statistical sampling to prove liability in this case. Sampling is a widely accepted method of proof in cases brought under New York law, including in cases relating to RMBS and involving repurchase claims. See Syncora Guarantee Inc. v. EMC Mortg. Corp., No. 09 Civ. 3106, 2011 WL 1135007, at *4 (S.D.N.Y. Mar. 25, 2011); MBIA v. Countrywide Home Loans, Inc., 30 Misc.3d 1201(A), 2010 WL 5186702 at *4 (N.Y.Sup.Ct.2010). Although Flagstar argues that the fact determination of material breach in any given instance requires consideration of an entire loan file renders the loans ill-suited to proof by statistical sampling, this argument is unpersuasive. The very purpose of creating a representative sample of sufficient size is so that, despite the unique characteristics of the individual members populating the underlying pool, the sample is nonetheless reflective of the proportion of the individual members in the entire pool exhibiting any given characteristic. Because the Court accepts sampling as an appropriate method of proof in this case and because it largely adopts Walzak's findings of material defects, the Court finds that the loans underlying the Trusts here at issue pervasively breached Flagstar's contractual representations and warranties.

Having found material breaches of the relevant contracts, the Court must determine whether Flagstar was made aware of the existence of these pervasive material and adverse breaches, and, if so, at what point in time. This is because § 2.04(d) of the SSAs provide that Flagstar's cure or repurchase obligations are triggered by Flagstar “becoming aware of” a breach. Pl. Ex. 198 § 2.04(d).

Flagstar argues that only actual awareness of material and adverse breaches will meet this requirement, as it claims that the Transaction Documents do not contemplate constructive notice. See, e.g., Pl. Ex. 198 § 2.04(c) (referring to “a breach,” rather than “breaches”); Pl. Ex. 198 § 2.04(e) (requiring Flagstar to accept transfer of a “Mortgage Loan as to which a breach has occurred”). Thus, according to Flagstar, Flagstar's knowledge as to a subset of breaches—whether that knowledge is based on Assured's repurchase demand or even its proof at trial of material breaches within the loan sample—is insufficient to create an obligation to repurchase any loans not included in that subset.

However, in its opinion denying Flagstar's motion to dismiss Assured's claims, the Court found that, for purposes of that motion, Assured, by informing Flagstar “ ‘of pervasive breaches' affecting the charged off loans” with its January 2009 repurchase demand, “rendered Flagstar constructively ‘aware’—or, at a minimum, put Flagstar on inquiry notice—of the substantial likelihood that these breaches extended beyond the charged off loan population and into the broader loan portfolio.” Am. Mem. at 18, ECF No. 56 (Oct. 27, 2011). The Court adheres to that decision now and finds that Assured's repurchase demand was sufficient to trigger Flagstar's obligations under § 2.04(d).

Although Flagstar claims that it “cured or rebutted” all of Assured's demands arising from the January 2009 repurchase demand, Flagstar's response denying the need to repurchase any loans cannot be sufficient to absolve Flagstar of awareness that Assured was claiming pervasive breaches of the representations and warranties. If that were so, then the 90–day cure-or-repurchase period could never begin to run even on the specific loans included in the demand, as Flagstar could deny awareness of breaches by claiming to refute Assured's allegations. To the extent that Flagstar argues that it must be notified as to each loan as to which a material breach is claimed with sufficient specificity to allow Flagstar to identify the loan and investigate the alleged breach, this requirement inappropriately places the burden of notification on Assured, when, as discussed above, the Court has found that Flagstar's responsibilities are triggered merely by awareness.²⁰ Thus, the Court finds that Flagstar was made aware of Assured's claim of pervasive and material breaches of the representations and warranties and therefore has breached its contract with Assured by failing to “repurchase or cure” the materially defective loans in the 2005–1 and 2006–2 securitizations, as required by the Transaction Documents.

20 The Court further notes that Assured has presented evidence that Flagstar may have been aware of problems in its underwriting at the time that the parties entered into these transactions, as Flagstar's auditor at that time, Wetzel Trott, found underwriting-related problems in the sample of loans it reviewed. See, e.g., Pl. Ex. 132, 133. Moreover, the Court notes that Flagstar also serviced the loans in the securitizations through 2010, so that, at least in some cases, it had information that would have suggested fraud or misrepresentations in the original loan application. For example, for the loan ending in 4977, Flagstar's servicing notes showed a conflict between the borrower's representation in the loan application that he was the president of Regional Financial Group and the borrower's hardship claim to Flagstar, claiming that he had been laid off as a police sergeant and loan broker. See Trial Tr. 1274:12–1278:8.

DAMAGES

The only remaining issue is the question of damages. This Court previously held “Flagstar's ‘cure or repurchase’ obligation to be the exclusive remedy available to [Assured] for Flagstar's breach of a representation or warranty.” Am. Mem. at 12, ECF No. 56 (Oct. 27, 2011) (citing Pl. Ex. 451 § 3.02(c); 198 § 2.04(e)). Thus, the contractual mechanism under which Assured is entitled to damages would require the re-transfer of the defective mortgages back to Flagstar, and a payment by Flagstar into each Trust in the amount of the “Transfer Deficiency” for that loan, as calculated by the Indenture Trustee. Pl. Ex. 198 § 2.04(d); see also Am. Mem. at 20, ECF No. 56 (Oct. 27, 2011) (determining that “the proper measure of damages for a charged off loan is the amount of the outstanding principal balance on that loan at the time it was ‘charged off’ by Flagstar”). Therefore, Assured is not entitled to direct payment of the amounts Flagstar should have paid for a repurchase, see Pl. Ex. 198 §§ 2.04(e), 2.07(a), 3.02(b), but rather to reimbursement of the claims it has paid to the bondholders to the extent that the amounts Flagstar should have paid into the Trust would be sufficient to cover Assured's claim payments.

As an initial matter, Flagstar challenges whether the use of sampling to prove damages inherently conflicts with the cure-or-repurchase remedy to which Assured is limited. Flagstar argues that sampling does not identify specific breaches in specific loans in the overall pools such that Flagstar can cure the defects as provided for in the Transaction Documents, nor can extrapolation identify the specific loans which Flagstar would accept for retransfer should breaches be unable to be cured. Thus, argues Flagstar, awarding damages to Assured here would deprive Flagstar of its side of the contractual repurchase bargain.

However, as to Flagstar's right to cure any deficiencies on the loans, this Court has already held that the 90–day cure period “has long since expired.” Am. Mem. at 20, ECF No. 56 (Oct. 27, 2011). Moreover, since, as Dr. Mason testified, Assured's damages model is based only on defective, defaulted loans, Flagstar's right to cure a breach is irrelevant because it is impossible to entirely cure any breach as to a mortgage loan that has already defaulted. Flagstar would also receive nothing back on defaulted loans, even if specific loans in the sample were identified and re-transferred. Thus, the Court rejects Flagstar's objections to sampling as a method of proving damages.

Flagstar also suggests that it in fact owes no damages to Assured on the 2005–1 Trust because current projections suggest that Assured will be reimbursed in full at the end of the transaction. See Trial. Tr. 815:1–17, 1137:5–12. Stanley Jursek testified at trial that Flagstar retained in the securitizations a residual interest, known as the “Transferor's Interest,” that entitles it to all remaining amounts left in the securitizations after all of the other parties to the transactions— e.g., the bondholders, the Trustee, and Assured—receive the payments to which they are entitled, including reimbursement of all claims paid by Assured. See JPPO ¶¶ 36–37; Trial Tr. 811:23–812:4. According to Jursek, to the extent that the net present value of Flagstar's Transferor's Interest in a securitization is positive, Assured is predicted to be repaid in full, and Flagstar expects to receive some value at the close of the transaction. Trial Tr. 817:19–22. As of June 2012, Flagstar estimated the net present value of its Transferor's Interest in the 2005–1 transaction at $7.6 million, but it predicted that there would be no remaining Transferor's Interest at the end of the 2006–2 transaction. Def. Ex. BAS; Trial Tr. 816:8, 817:19–22. At trial, Flagstar also introduced the testimony of Ann Rutledge, its structured securities valuation expert,²¹ who modeled the cash flow of the 2005–1 transaction and concluded that Flagstar's Transferor's Interest would have a positive value at termination of $1.4 million, indicating that Assured would be reimbursed in full for claims paid on the 2005–1 transaction. Trial Tr. 1129:10–23, 1137:5–12. Rutledge offered no opinion on the 2006–2 transaction. Id. at 1139:22–23.

21 Rutledge is the Chief Strategist for R & R Consulting, which provides valuation and structure services for structured securities. Trial Tr. 1126:21–1127:1. Rutledge holds a master's degree in business administration and began working at Moody's Investor Service in 1995 as a structured finance analyst. Id. at 1127:14–20.

This Court held in its summary judgment decision that “[t]here is no provision in the contract that requires Assured to wait until all activity on the transactions is complete before obtaining any recovery. Moreover, if the fund does recover more in payments than Assured has predicted, Assured will be responsible for paying this money to Flagstar.” Mem., 892 F.Supp.2d at 606. The Court continues to adhere to its previous position. What is more, at trial both Jursek and Rutledge acknowledged that these amounts are not certain: Flagstar's estimate of its Transferor's Interest dropped twenty percent in the six months between December 2011 and January 2012 (from $9.5 million to $7.6 million), see Trial Tr. 859:4–22, and Rutledge reluctantly acknowledged that the rate of error on her model was unknown, as it was not generally used as a forecaster, see id. at 1141:13–20. Thus, the Court refuses to allow the damages owed to Assured on the 2005–1 transaction to remain in limbo while hoping for a positive outcome down the road, and therefore rejects this challenge to an award of damages. See Tractebel Energy Mktg., Inc. v. AEP Power Mktg., Inc., 487 F.3d 89, 112 (2d Cir.2007) (“The risk that the future might reveal the district court's assumptions to be false is appropriately borne by ... the breaching party.”).

Moving on to the calculation of damages, what Assured showed through Dr. Mason's damages model was that Flagstar's failure to repurchase those defaulted loans that had breached the representations and warranties directly and proximately caused Assured to improperly bear the burden of paying claims on the transactions. Dr. Mason testified that, had Flagstar repurchased the defective loans, Assured would have been reimbursed for all the claims paid to the bondholders. Trial Tr. 428:4–16; see also Diesel Props S.r.l. v. Greystone Bus. Credit II LLC, 631 F.3d 42, 52–53 (2d Cir.2011) (“Causation is an essential element of damages in a breach of contract action; and, as in tort, a plaintiff must prove that a defendant's breach directly and proximately caused his or her damages.” (quoting Nat'l Market Share v. Sterling Nat'l Bank, 392 F.3d 520, 525 (2d Cir.2004))). The Court discounts to some small degree Assured's claimed breach rate based both on the Court's concerns, discussed above, regarding certain aspects of Walzak's analysis and on the errors in Dr. Mason's calculations ( i.e., that some of the loans treated as charged off were not in fact sufficiently delinquent to be deemed charged off, see Trial Tr. 460:4–6, 465:2–3, 466:1–2). Nonetheless, the Court finds that Assured has sufficiently proven that it should be reimbursed in full for all claims paid on the underlying transactions, as the Court accepts Dr. Mason's largely unrebutted testimony that Flagstar would be obligated to reimburse Assured for the entirety of the claims Assured paid under a wide range of defect rates. Id. at 429:14–15.

Dr. Mason testified that, in total, Assured has paid about $14.7 million in claims arising from the 2005–1 Trust and approximately $75.4 million in claims arising from the 2006–2 Trust, for a total of approximately $90.1 million in claims. Pl. Exs. 458 at 4, 459 at 5. At trial, Assured did not contest that Flagstar had reimbursed Assured in the amount of about $950,000. See Trial Tr. 446:20–447:24. Therefore, subtracting the amounts Flagstar already paid (which Dr. Mason admitted were left out from his damages calculation), the Court finds that Assured is entitled to collect $89.2 million in damages, plus interest. SeeN.Y. C.P.L.R. §§ 5001, 5002 (allowing for interest in breach of contract actions).

There was debate at trial as to the proper interest rate to apply to reimbursements due to Assured. Section 6.03 of the I & Is provides that payments made to Assured pursuant to the I & I “shall bear interest at the Late Payment Rate from the date when due to the date paid.” Pl. Ex. 90 § 6.03. The Late Payment Rate amounts to the prime rate plus two percent. Pl. Ex. 90 App. I at 2. In his calculation of damages, Dr. Mason applied the Late Payment Rate to claims and paymentsmade in the normal course, but applied the statutory prejudgment interest rate (nine percent) from the date Assured should have been reimbursed and to those claims that Assured should never have paid, had Flagstar complied with its contractual obligations. Trial Tr. 431:1–18. Given the background rule that contractual rates of interest supersede New York's statutory interest rate, see Nuera Commc'ns, Inc. v. Telron Commc'ns USA, Inc., 00 Civ. 9167, 2002 WL 31778796 (S.D.N.Y. Nov. 15, 2002), and the I & Is statement that the Late Payment Rate governs “to the date paid,” the Court finds that the Late Payment Rate is the appropriate interest rate to be applied in calculating prejudgment issues, as of the dates on which Assured made its claims payments.

Finally, after trial, the Court received post-trial briefing from the parties on the issue of whether Flagstar is contractually obligated to reimburse Assured for its costs and expenses in relation to Assured's initial repurchase demands and its prosecution of this lawsuit. Section 3.03(b) of the I & Is requires that Flagstar reimburse Assured for

any and all charges, fees, costs and expenses that FSA or its affiliates may reasonably pay or incur, including, but not limited to, attorneys' and accounts' fees and expenses, in connection with ... (iii) the administration, enforcement, defense or preservation of any rights in respect of any of the Transaction Documents, including ... participating in any litigation ... relating to any of the Transaction Documents ... or the Transaction....
Pl. Ex. 90 § 3.03(b).

As an initial matter, the Court finds that awarding costs and fees would not be contrary to its prior ruling that Assured was limited to the contractual cure-or-repurchase remedy as its “sole remedy,” notwithstanding Flagstar's arguments to the contrary. See Am. Mem. at 10–12, ECF No. 56 (Oct. 27, 2011). The Court's opinion granting in part Flagstar's motion to dismiss did not directly address section 3.03 of the I & Is, while it did address the interplay of other contractual terms, and for that reason alone, the opinion does not preclude the award of costs and fees to Assured. Neither is the logic of the Court's opinion necessarily contrary to awarding costs and fees. The Court stated in its opinion that “the ‘sole remedy’ provisions of the Transaction Documents do not preclude AGM from bringing suit against Flagstar in the event that, as alleged here, Flagstar refuses to comply with its repurchase obligations.” Am. Mem. at 12. If the “sole remedy” provisions do not preclude Assured from bringing suit, then it seems illogical that that remedy would preclude Assured from enforcing its collateral rights to reimbursement for the costs and expenses of that litigation. Thus, the Court finds that Assured is entitled under the I & Is to reimbursement for its reasonable fees and costs in this litigation and the repurchase demands that precipitated it.

Assured claims reimbursable fees and expenses in the amounts of $3.69 million in attorneys' fees, $443,000 in vendors' fees, and $1.01 million in expert fees, for a total of $5.14 million. Based on its decision to withdraw its servicing claims prior to trial, Assured reduced its attorneys' fees by 10% and the fees of its experts who worked on servicing-related issues by 25%. Assured's counsel also calculated what it deemed to be a reasonable amount of attorneys' fees not based on its actual partial-contingent fee arrangement with Assured, but rather on a lodestar method that it claims is more reasonable to Flagstar. Because the Court is unable to assess the validity of Assured's counsel's claims that its servicing-based reductions are sufficientand that its fee calculations are reasonable without narrative descriptions of its own and its experts' work and the terms of its agreement with its client, the Court orders that Assured provide the documentation underlying its claim (for in camera review where necessary) by February 15, 2013. See New York State Ass'n for Retarded Children, Inc. v. Carey, 711 F.2d 1136, 1148 (2d Cir.1983) (requiring that attorneys document their fee applications with records specifying “the date, the hours expended, and the nature of the work done”).

In conclusion, the Court hereby grants judgment in favor of Assured on its claims for breach of contract against Flagstar in the amount of $89.2 million plus contractual interest and attorneys' fees and costs to be determined hereafter.

SO ORDERED.