The news gets worse for Bank of America. Not only will it have to eat massive numbers of Countrywide-originated loans, but the bank may have to complete repurchases sooner than previously thought. Judge Eileen Bransten’s long-awaited evidentiary ruling in the New York state court makes it clear – MBIA can use statistical sampling to prove its claims against Countrywide/BofA. The ruling provides the bond insurer–and other insurers and private label investors–with a short cut to proving claims against lenders and originators of defective mortgage loans.
Bransten’s Order has already had an impact in newly-filed litigation over the losses associated with private label residential mortgage backed securities (RMBS). A $700 million lawsuit filed by Allstate Inusurance Co. against Bank of America this past week refers directly to language from Bransten’s Order (available here and embedded below).
In her 15-page Order, Bransten found that MBIA will “be allowed to use and present evidence for its case through statistical sampling” (p. 15). This means that MBIA will not have to present evidence of fraud or breach of contract for each of the 300,000-plus loans at issue in its case to a judge or jury; instead, MBIA will be able to evaluate a much smaller but statistically significant sample of loans and extrapolate the findings to the rest of the loans in the challenged securitizations.
In the course of making her ruling and rejecting the vigorous arguments made by Countrywide in opposition to MBIA’s motion, Judge Bransten found that the use of statistical sampling of large populations was not novel, was generally accepted in the scientific community, and was appropriate in the case at bar. Bransten also explicitly held that her decision, “has the possibility of saving the parties and the court from significant litigation time and may significantly streamline the action without compromising either party from proving its case” (p. 13).
The outcome of MBIA’s evidentiary motion (also known as a motion in limine) was not entirely unexpected based on Judge Bransten’s prior comments. In a June 16, 2010 hearing (transcript available here), Bransten said,
I think that it makes all the sense in the world that you can use a sample to prove the case because otherwise I can’t imagine a jury listening to 386 thousand cases. Even if you have that available, nevertheless you are not going to present that to a jury or even a judge. I’m patient but not that patient. So, therefore it is going to be a sample in the end…
Yet the fact that an order in this regard has officially been put to paper–and in a closely-watched case such as this one–is already having a major impact in the MBS litigation world. On Monday, Countrywide and Bank of America were sued by Allstate Insurance Co. and its affiliates over $700 million of RMBS purchased by the insurer (complaint available here). In that suit, Allstate proposed to use samples of 1600 loans (800 defaulted loans and 800 randomly-sampled loans) from each of 14 securitizations (there are 61 securitizations at issue) to prove its claims. In support of this methodology, Allstate asserted that,
Allstate‘s sample sizes of Mortgage Loans are more than sufficient to provide statistically-significant data to demonstrate the degree of misrepresentation of the Mortgage Loan characteristics. Analyzing data for each Mortgage Loan in each Offering would have been cost-prohibitive and unnecessary. Statistical sampling is an accepted method of establishing reliable conclusions about broader data sets, and is routinely used by courts, government agencies, and private business. As the sample size increases, the reliability of its estimations of the total population increase as well. Experts in RMBS cases have found that a sample size of just 400 loans can provide statistically significant data, regardless the size of the actual loan pool, because it is unlikely that so large a sample would yield results vastly different from results for the entire population. (Allstate Complaint, p. 38)
Filed by Quinn Emanuel, the same law firm representing MBIA and other monolines, the Allstate Complaint is novel for its assertion of a “matching strategy” by Countrywide–that is, that Countrywide was willing to approve any mortgage feature offered by its competitors, resulting in Countrywide “mixing and matching the worst features of mortgage products from different competitors,” and creating a product that was “very aggressive within the industry” (Allstate Complaint, p. 2). Such a strategy required Countrywide to systematically abandon its guidelines and resort to the widespread use of unsupported exceptions, according to the Complaint.
As the Allstate Complaint shows, Bransten’s decision in the MBIA case has already dealt a blow to banks trying to defend themselves against mortgage putback liability, as it significantly reduces the prospective litigation costs, as well as the projected timeline, for investors seeking to compel originators to buy back defective loans with the help of the courts. However, the decision is particularly harmful for BofA’s financial outlook and the credibility of its executives. In addition to shorting the path to court-mandated repurchases by the bank, which has the most potential putback liability of any of its peers based on its acquisitions of Countrywide and Merrill Lynch, the decision undermines several statements made by BofA’s CEO, Brian Moynihan, regarding his strategy for the battle over rep and warranty liability. For example, during BofA’s 2010 Q3 earnings call, Moynihan stated that, “This really gets down to a loan-by-loan determination and we have, we believe, the resources to deploy against that kind of a review… we will go in and fight this. It’s worked to our benefit to–we have thousands of people willing to stand and look at every one of these loans.”
Moynihan has also stated that BofA would approach this type of litigation like “hand-to-hand combat,” disputing individual putbacks and dragging out litigation as long as possible to allow the bank’s earnings to offset these contingent liabilities. Bransten’s Order presents a significant roadblock to the execution of that strategy. Now, instead of going loan-by-loan, MBIA will be able to use a surprisingly small sample–less than 1.6% of the total loan population in dispute–to prove incidences of breach or fraud in the entire pool.
In its motion in limine, MBIA proposed using a statistical sampling methodology that included sampling 400 loans from each of the 15 securitizations at issue; stratifying the samples into mutually exclusive subgroups based on the characteristics of the borrower’s credit score, combined loan to value ratio (CLTV) and Countrywide’s documentation program; and then dividing each of these subgroups into further subgroups to ensu
re that important loan characteristics were adequately represented in the samples. The monoline also proposed using delinquency status to stratify the samples used to prove its servicing contract and implied covenant claims. MBIA asserted that this methodology would provide a confidence level of approximately 95% with a 5% margin of error. Though Bransten ruled that MBIA could use this methodology to present evidence at trial, she stopped short of rejecting Countrywide’s arguments that this methodology was flawed and subject to challenge, arguments that she conceded “are not without merit” (p. 11). Instead, she ruled that those challenges were “premature,” and that “Defendant’s cited issues will be decided by the trier of fact as pertaining to the weight, rather than the acceptability, of the evidence” (p. 12).
Bransten’s commentary signals that MBIA may not be out of the woods yet in proving its case against Countrywide. As an initial matter, the sample size seems exceedingly small compared to the overall loan population. Due diligence firms like Clayton, who were often hired by investment banks to perform due diligence checks on loan pools before the bank would buy them, often took a 10-20% sample to ensure adequacy. As recounted above, Allstate has chosen a sample size of 1600 loans–four times the sample size chosen by MBIA. With the cost of hiring a third party firm to review a loan file averaging around $300-350, meaning that it would cost MBIA over $5 million more to review a sample of 1,600 loans across its 15 securitizations, it’s easy to see why MBIA is trying to keep its sample size small. Still, while I’m no statistician, it seems like cutting corners in this way could open MBIA’s data up to a number of technical challenges.
Furthermore, it’s not at all clear how the representative sample would be applied to the overall loan population, even if it was established to be representative by the trier of fact. The contractual remedy for proving that Countrywide breached a rep and warranty with respect to a loan is the repurchase of that particular loan. How would that remedy be applied to the overall loan population if MBIA’s sample showed, for example, that 60% of loans breached at least one rep and warranty? In other words, which loans would Countrywide be forced to repurchase, and how would the court ensure that size and the loss severity of those loans lined up with those found to be defective in the sample?
Still, this decision is a definitive win for MBIA and all RMBS investors hoping to recoup some portion of their losses. Now that MBIA has received the loan files underlying the securities at issue, it’s only a matter of time before it will be able to find significant percentages of loans with (often multiple) material deficiencies. By enabling the insurer to project findings on such a small sample to the rest of the pool, this decision provides MBIA with a monumental shortcut to establishing Countrywide’s repurchase liability. Only time will tell if this changes BofA’s strategy of hand-to-hand combat and brings the bank to the negotiating table, or simply provides the bank with one more issue on which to put up a fight.
[Thank you, as always, to Manal Mehta for sharing his perspective and real time updates on this case – IMG]
Bransten Sampling Order (MBIA v. Countrywide) http://d1.scribdassets.com/ScribdViewer.swf?document_id=46008294&access_key=key-2bvsg4dsfur86xdm5ujv&page=1&viewMode=list
