The Rise and Fall of Statistical Sampling in RMBS Cases

The 2008 mortgage crisis prompted a wave of residential mortgage-backed securities and repurchase litigation as trustees, certificate holders (i.e. investors), monoline insurers, securitizers and other stakeholders pursued claims related to loans securitized in RMBS trusts. One of the hallmarks of large-scale RMBS litigation is the large number of loans in dispute. These cases usually include allegations relating to thousands (or tens of thousands) of loans. In order to reduce the burden and expense of proving such claims on a loan-by-loan basis, plaintiffs typically have sought to prove their allegations through the use of statistical sampling. In other words, rather than allege individual breaches on thousands of loans, plaintiffs will hire a statistics expert to draw a sample, engage a re-underwriting expert to determine the proportion of the sample loans in breach, and then hire yet another expert to extrapolate the results to the overall population. In the early days of RMBS litigation, this strategy was highly successful. However, in recent years, a new trend has emerged. Courts have grown increasingly skeptical about the use of statistical sampling in repurchase cases, especially where language in the applicable loan sale agreements requires a loan-level determination of breach in order to trigger a repurchase obligation. This article discusses the growing trend in courts towards rejecting the use of sampling in such cases and the implications of such trends.

Originally published in Law360; reprinted with permission.