In a prior blog post, we discussed a decision (reported at 2014 WL 1282293) denying Apple’s motion to exclude the opinion of plaintiffs’ antitrust economics expert, Dr. Roger G. Noll (“Noll”), filed in support of plaintiffs’ motion for class certification.
In this post, we’ll review the same judge’s decision (reported at 2014 WL 1282298) granting plaintiffs’ motion to exclude the opinions of Apple’s experts, Dr. Joseph Kalt and Mr. Jonathan Orszag, filed in opposition to class certification.
As discussed in the prior blog post, the Court found that Apple and leading publishers had devised a new model – the “agency” model -- for e-book sales under which retailers such as Apple would act as agents to sell e-books at prices fixed by the publishers. After signing agreements with Apple, the publishers forced Amazon to adopt the new model as well. As a result, e-book prices jumped dramatically as publishers raised prices to the caps set under agreements with Apple and Amazon.
In August 2011, several class actions were filed alleging that the price fixing agreements between Apple and five leading publishers violated antitrust laws. The DOJ and various states filed their own lawsuits. The publishers all settled with the plaintiffs, but Apple elected to fight on. After a bench trial, Apple was found liable for violating Section 1 of the Sherman Antitrust Act, and the court scheduled a damages trial.
Plaintiffs’ expert, Dr. Noll, had run a multivariate regression analysis that determined what the price for given e-books would have been “but for” the illegal price-fixing, and proposed a measure of damages equal to the difference between the “but for” prices and the actual, sharply higher e-book prices that were charged after the “agency” model kicked in.
Apple’s experts did not conduct their own rigorous statistical analysis of factors affecting pricing and damages. Instead, they primarily tried to poke holes in Noll’s regression analysis. The failure of these experts to support their own opinions with adequate analysis proved to be their undoing (with one exception).
Kalt argued that Noll’s model was unreliable because it failed to take into account numerous factors and phenomena that Kalt claimed exist in the e-book market. The Court rejected all of Kalt’s criticisms. The statistical analysis in the decision becomes quite complex at times, but in a nutshell, the Court found that, in most cases, Kalt had not conducted rigorous analysis of his own in support of many of his opinions, had expressed opinions contradicted by the record evidence, and/or relied on demonstrably false assumptions.
For example, the Court found that Kalt’s claim that the price of 60% of e-books sold dropped following the adoption of the agency model, was flawed due to his mislabeling of a significant number of “post-agency” prices as “pre-agency” prices (thus creating the illusion of a drop). Insofar as Kalt’s pricing analysis was based on demonstrably false assumptions, the Court ruled it inadmissible.
Similarly, Kalt’s claim that book prices during the relevant periods showed significant variance due to “churning” and “dispersion” was also rejected. Instead, the Court found a remarkable uniformity in the price increases that followed the introduction of the agency model. While there was some variance not explained by Noll’s model, the Court concluded that “plaintiffs do not have the burden of showing that all e-book prices behaved in just the same way.”
The Court further dismissed Kalt’s arguments that Noll’s model suffered from millions of “false positives,” and failed to take into account “buzz” from celebrity endorsements and other unusual influences. In each case, the Court found that Kalt’s opinions were based on flawed analyses.
Orszag objected to Noll’s damages model on several grounds. First, he argued that Noll used an inappropriate control group when he included e-book publishers beyond the Big Six in his analysis. He also claimed that Noll used incorrect time periods. Finally, Orszag contended that Noll’s calculation of a final damages figure failed to account for benefits that consumers received due to the price-fixing conspiracy.
The Court dismissed Orszag’s criticisms concerning the purported benefits of the conspiracy on the ground that such benefits were outside the relevant “market” insofar as they related to different products such as pricing of Amazon’s Kindle. Other benefits cited by Orszag were deemed speculative.
However, the Court found admissible Orszag’s criticisms regarding Noll’s definition of the control group and time period for his regression analysis. Perhaps not surprisingly, the Court held Orszag’s opinion on these points admissible because Orszag undertook the time and effort to re-run Noll’s regression analysis using the control group and timeframe that he advocated.
As noted, the statistical analysis and terminology in the Court’s opinion becomes extremely complex at times. However, one critical lesson emerges – experts need to do their homework. That is, experts cannot successfully poke holes in the other side’s experts if they don’t support those criticisms with rigorous analysis on par with the analysis of the experts being criticized. Extensive due diligence and preparation is clearly key to winning the battle of the experts.
We’d be interested to hear from readers concerning “war stories” preparing and/or cross-examining experts in a complex field such as econometrics.