U.S. Bank National Association v. Goldman Sachs Mortgage Company

Background

U.S. Bank National Association serves as trustee for two trusts that pool mortgage loans backing residential mortgage-backed securities. Goldman Sachs Mortgage Company sponsored the trusts and acquired approximately 8,500 loans worth approximately $1.7 billion. GS Mortgage Securities Corp. deposited the loans into the trusts. The parties’ agreements included representations and warranties about the loans and required Goldman Sachs Mortgage Company to cure or repurchase loans affected by qualifying breaches.

U.S. Bank received investor notifications identifying at least 617 allegedly defective loans in the GSAMP Trust 2007-HE1 and 1,041 allegedly defective loans in the GSAMP Trust 2007-HE2. U.S. Bank notified Goldman and demanded that it cure or repurchase the identified loans and other defective loans. According to the opinion, Goldman had not cured any breaches or repurchased any loans when the demands were made. U.S. Bank sued for breach of contract and sought specific performance and damages.

The proposed sampling method

U.S. Bank moved for partial summary judgment on whether it could use statistical sampling to prove liability and damages at trial. Its proposed method involved randomly selecting 400 loans from each loan group in each trust, for 800 loans per trust and 1,600 loans total. The selected loans would come from liquidated loans—that is, loans that had defaulted and been charged off—with at least $100 in losses. U.S. Bank proposed reviewing active loans individually.

Its experts would re-underwrite the selected loans by comparing each loan’s borrower, loan, property, and documents to the contractual warranties. They would determine which sampled loans had material breaches that adversely affected value, calculate a breach rate, and extrapolate that rate to the larger loan groups. U.S. Bank said this approach would avoid re-underwriting approximately 2,700 loans and shorten discovery by about 20 months. The opinion states that the reliability and admissibility of the experts’ method under Federal Rule of Evidence 702 were separate issues for a later motion.

Court’s analysis

The court applied New York contract law. It explained that summary judgment on a contract claim is proper only when the relevant contractual language is wholly unambiguous and has a definite meaning. The contracts did not expressly address statistical sampling, but the court concluded that their provisions described a loan-specific process.

The agreements referred to a breach affecting “such” a mortgage loan, required cure or repurchase of the affected mortgage loan, and required a determination of whether the breach materially and adversely affected the value of the particular loan. The repurchase price also depended on individual loan characteristics, including unpaid principal balance, mortgage interest rate, servicing advances, and expenses. The court therefore found that the contracts consistently used singular, loan-specific terms for breach, materiality, remedy, and damages.

The court rejected U.S. Bank’s argument that limiting sampling to liquidated loans changed the analysis. It reasoned that the issue was not only whether a loan would be transferred back to Goldman, but also whether breach, materiality, and the repurchase price could be established without loan-specific proof. The court also considered conflicting decisions from other courts, state-court decisions approving sampling, the parties’ concerns about expense and delay, and the possibility that the experts’ methods could be reliable. It concluded that those considerations did not establish that the contracts unambiguously permitted sampling.

The court did not decide Goldman’s alternative argument that U.S. Bank should be barred from advocating sampling based on positions it had taken in earlier cases. It also did not resolve disputes about whether U.S. Bank’s notice satisfied the contracts’ requirements or whether those requirements were prerequisites to suit.

Disposition

Judge Alison J. Nathan denied U.S. Bank’s motion for partial summary judgment. The court also denied the parties’ requests for oral argument. It ordered the parties to file a revised case-management plan within two weeks. The opinion does not finally decide whether statistical sampling will be admissible at trial or resolve the underlying breach-of-contract claims.