Statistical Sampling and Individual Liability: FCA Lessons Learned from Whopping $298.5 Million Verdict in United States v. Americus Mortgage Corp.
Imposing civil penalties and treble damages under the False Claims Act (“FCA”), the Southern District of Texas recently denied defendants’ post-trial motions and entered a staggering $298,498,325 verdict against two mortgage companies and their CEO. In United States v. Americus Mortgage Corporation, et al., No. 4:12-cv-2676, (“Allied Mortgage”) the government sued Allied Home Mortgage, its various entities, and its CEO Jim Hodge for submitting false claims to HUD and perpetrating housing fraud over a ten-year period. Apart from the CEO’s individual liability, this litigation is noteworthy because the court sided with the government in two important post-trial rulings on proximate causation and statistical sampling in large-scale fraud cases. The sheer size of this verdict reinforces the importance for companies—and their executives—to carefully follow the trends in FCA litigation to avoid potentially staggering liability.
The Risk of CEOs’ Individual Liability Is Real.
In Allied Mortgage, the United States sought damages under the FCA and FIRREA for losses stemming from defaulted housing loans. The government alleged that Allied violated HUD regulations, provided false statements to FHA, and submitted false claims for reimbursement to HUD for fraudulent loans in default. The government presented documents and testimony establishing that Hodge knew of, and approved, Allied’s business practices and personally directed subordinates to submit false quality control reports to HUD. The jury held Hodge jointly liable for over $23 million.
The court had no trouble rejecting Hodge’s post-trial motion to set aside the verdict, finding the evidence supported the government’s case. Businesses should take heed: the potential for crippling liability extends to executives with knowledge and operational control.
The Government’s Proximate Causation Evidence Proved a “Sufficient Connection” to the Government’s Loss.
Defendants also challenged the government’s case on the basis that it did not prove the alleged poor underwriting practices proximately caused any loan defaults. The court rejected defendants’ argument, and the proximate cause standard the court articulated raises the question of how stringent that standard actually is.
The court explained that proximate causation “is not so stringent as to require elimination of all alternative possible causes” and that an event is the proximate cause if it has “a sufficient connection to the result.” It found the government’s damages were a foreseeable consequence of defendants’ fraudulent conduct and described the chain of causation as follows: defendants originated loans from unregistered branches, skirting HUD’s regulatory requirements; underwriters issued false statements about borrowers’ creditworthiness; this conduct increased the risk of borrower default; and in fact, borrowers defaulted at a high rate, leading to the government’s losses. Notably, the court did not consider the sheer number of events in the causal chain or the fact that one important event—the borrowers’ default—can be caused by myriad factors other than poor underwriting.
The key takeaway here is that court essentially equated proof of increased probability of default with proof that poor underwriting—as opposed to other factors—caused the losses in these cases.
The Government’s Statistical Extrapolation Method Proved Damages.
Over defendants’ protests, the United States engaged experts to draw random samples from over 17,000 loans the defendants submitted to FHA between 2001 and 2011. The experts analyzed these samples and then extrapolated the findings to deduce the total number of falsely underwritten loans, concluding that defendants collectively submitted 1,295 unlawful insurance claims. The defendants claimed that the government should have been required to prove that defendants proximately caused each individual default rather than merely drawing that conclusion from a sample data set.
The court flatly rejected the defendant’s position, but its conclusion raises several issues. For one thing, while not argued by defendants here, permitting a jury to hear a damages calculation based on extrapolation methodology rather than individual findings raises substantial due process concerns. In fact, the defendants in an earlier Fourth Circuit case, United States ex rel. Michaels and Whitesides v. Agape Senior Community, Inc., argued that sampling would deprive them of their constitutional right to a jury trial on each individual alleged false claim. The Fourth Circuit declined to address this argument, but the very same concerns are raised here.
Second, the court’s application of this methodology in the FCA context rested on the proposition that courts routinely use extrapolation methodology in cases involving large amounts of data, but that point is far from settled. The primary Fifth Circuit case the district court cited in support of that point, In re Chevron U.S.A., Inc., 109 F.3d 1016 (5th Cir. 1997), is not an FCA case at all. But more importantly, the district court’s opinion here in Americus Mortgage does not cite or analyze a later Fifth Circuit case questioning Chevron on the use of extrapolation methodology. In Cimino v. Raymark Indus., Inc., 151 F.3d 297 (5th Cir. 1998), the court rejected the use of statistical sampling in mass tort litigation and declined to read Chevron as endorsing extrapolation as the district court did here. Id. at 318.
To be sure, the district court’s failure to analyze or even cite Cimino compromises its analysis on this point. We anticipate that defendants will continue to dispute this issue as more courts grapple with the use of extrapolation in FCA litigation.