Testing the Reliability of an Expert's Predictive Model

It appears that the United States District Court, Eastern District of Virginia, has broadly interpreted Federal Rule of Evidence 702 to now include testing the reliability of an expert's predictive model by comparing the results of the model to the actual market values during an "unaffected" time period. A recent case heard by this court in which the court ruled that the Plaintiff's expert's testimony should have been excluded under Daubert illustrated this very concept. See United States of America v. Birkhart Globistics GmbH & Co., et al., Civil Action Nos. 1:02-cv-1168 (AJT/TRJ) and 1:07-cv-1198 (AJT/TRJ) (E.D. Va. Dec. 24, 2014).

In Birkhart Globistics, the jury found that the defendants knowingly caused false claims to be submitted and awarded the plaintiff $33.6 million in damages. The defendants moved for judgment as a matter of law and the district court granted defendants' motion and vacated the damages award based on the following critique of the testimony provided by plaintiff's expert.

The testimony of the plaintiff's expert regarding damages was based on a regression model that sought to predict the prime rates for shipping goods overseas. During the trial, evidence was produced indicating the plaintiff's expert's regression model did not accurately predict the actual market prime rates during an "unaffected" period. For example, the expert's regression model did not produce a predicted prime rate equal to the actual prime rate during a period of time that no wrongful actions were alleged.

The district court concluded that plaintiff's expert failed to explain the inconsistencies between the actual market prime rates and the prime rates predicted using plaintiff's expert's regression model during an "unaffected" period. The court questioned the aspect of Federal Rule of Evidence 702 relating to whether the testimony was the product of reliable principles and methods. The court questioned not only why the predicted prime rates from the plaintiff's expert's regression model did not equate to the actual prime rates during an "unaffected" period but also why plaintiff's expert was unable to explain why the inconsistencies occurred.

The court concluded that if the regression model cannot accurately predict the prime rates during the "unaffected" period, the regression model could not be reliable to predict the "but for" prime rates during the "affected" period. Because the plaintiff's expert's damage calculation was based on the predicted "but for" prime rates, the court concluded that the plaintiff's expert's testimony should have been excluded under Daubert and, ultimately, vacated the jury's damages award.

Not only should experts be cognizant of the fact that expert testimony must be the product of reliable principles and methods, experts should also consider testing and comparing the results of any predictive model to actual market values during an "unaffected" period, when possible. If the expected results from the model are not consistent with the actual market values during the "unaffected" period the expert should be prepared to explain those inconsistencies.

Keywords: Birkhart Globistics, Daubert, commercial, business, litigation, expert, regression, model, predict, damages

_{Jason T. Wright and Richard M. Ruffing are with Stout Risius Ross, Inc., in Chicago, Illinois.}

_{Copyright © 2016, American Bar Association. All rights reserved. This information or any portion thereof may not be copied or disseminated in any form or by any means or downloaded or stored in an electronic database or retrieval system without the express written consent of the American Bar Association. The views expressed in this article are those of the author(s) and do not necessarily reflect the positions or policies of the American Bar Association, the Section of Litigation, this committee, or the employer(s) of the author(s).}